HABITSANDTECHNOLOGYFIT:ASTUDYOFTECHNOLOGY

ACCEPTANCE

Athesissubmittedinfulfillmentoftherequirementsforthe
degreeofDoctorofPhilosophy

LuisGerardoSnchezAcenjoCarrillo
(LegallyknowninAustraliaas:LuisSatch)

BachelorofArts(Communication)
MasterofBusinessAdministration


1













SchoolofManagement
CollegeofBusiness
RMITUniversity
March2014

ii

 DECLARATION

I certify that except where due acknowledgement has been made, the work is that of the

authoralone;andtheworkhasnotbeensubmittedpreviously,inwholeorinpart,toqualify

foranyotheracademicaward;thecontentofthethesisistheresultofworkwhichhasbeen

carried out since the official commencement date of the approved research program; any

editorial work paid or unpaid, carried out by a third party is acknowledged; and, ethics,

proceduresandguidelineshavebeenfollowed.

Signature:




LuisGerardoSnchezAcenjoCarrillo
(LegallyknowninAustraliaas:LuisSatch)

iii

 ACKNOWLEDGEMENTS

Undertaking a PhD is paradoxically the most solitary stage in anyones education; because
despite working alone for four years, there are numerous people and organizations
contributingbefore,duringandaftertheelaborationofathesisinawaythatasingleperson
canconductapieceofresearchandbeawardeddoctoraldegree.

TherelationshipwithsupervisorsisperhapsthemostinfluentiallinkforaPhDstudentduringa
candidature. Therefore, I shall say thank you first: to my supervisors Professor Adela
McMurray and Dr. Nuttawuth Muenjohn. I will cherish every minute spent together as their
guidancewassoundandwiseatalltimes.Iamgratefulthattheyshowedauthenticsupport,
andbelievedinme.

I would like to recognize and thank my wife, Elena, for her active support, love and
understanding during this wonderful journey of becominga researcher. After all she was the
closest at every step of this and many other worthwhile journeys. She shared, enjoyed and
maybe even suffered some of the unavoidable milestones of my studies... No doubt she
deservesmylove,gratitude,andrecognitiontoagreaterextentthanIcouldpossiblyexpressin
writtenwords.

A large amount of the credit for all I have accomplished has to be granted to my parents,
AdrianaandGerardo,astheyalwayscaredwithallthedeepimplicationsthatcaringbrings.I
wouldalsoliketoacknowledgemysister,Ingrid,because,asasourceofconstantcreativityand
novelty,sheplantedtheseedofthegreatideathatbroughtmetoAustralia.

Special acknowledgment for my studies should be given to RMIT University and CONACYT
whichprovidedtheacademicstructureandfinancialsupporttoshapemeasaresearcher;The
University of Melbourne was instrumental in providing a significant part of my statistical
training,andOptimalWorkshopswhichkindlysponsoredmyresearchbyprovidingaccesstoa
cloudsoftwarekeyofmyresearch.

Innumerable friends and scholars deserve a mention, and I am deeply thankful for each of
them. However, I must mention a few of their names as they intervened actively and/or
provided a crucial collaboration at some point of my doctoral studies; without their
contributionIwouldnotbeherecompletingaPhD,oratleastnotnow.Thankyou:Professor
Brian Corbitt, PhD; Dr. Carlos La Bandera, Associate Professor David Gilbert, PhD; Felipe
Velazquez Lpez, Dr. Jos Guadalupe Snchez Avia, Dr. Peter Chomley, Dr. Siddhi
Pittayachawan.

iv

 DEDICATION

Tothosewhoarecommittedtotransformtheworldintoabetterone

bychangingthemselvesfirst

 TABLEOFCONTENTS

DECLARATION..............................................................................................................................................iii
ACKNOWLEDGEMENTS................................................................................................................................iv
DEDICATION..................................................................................................................................................v
TABLEOFCONTENTS....................................................................................................................................vi
LISTOFTABLES..............................................................................................................................................x
LISTOFFIGURES...........................................................................................................................................xi
ABSTRACT....................................................................................................................................................xiii

CHAPTER1INTRODUCTION..........................................................................................................................1
1.1Objective............................................................................................................................................1
1.2Researchobjectives...........................................................................................................................1
1.3Background........................................................................................................................................2
1.4Justification........................................................................................................................................4
1.5Significance........................................................................................................................................6
1.6Researchquestions............................................................................................................................6
1.7ResearchMethodology......................................................................................................................7
1.8Structureofthethesis.......................................................................................................................9
1.9Keyconcepts....................................................................................................................................10
1.9.1Habittechnologyfit..................................................................................................................10
1.9.2Behavioralintention..................................................................................................................10
1.10UTAUTtheoreticalframework.......................................................................................................10
1.11Keycontributions...........................................................................................................................11
1.12Limitationsandfutureresearch.....................................................................................................13
1.13Summary........................................................................................................................................14
CHAPTER2LITERATUREREVIEW................................................................................................................15
2.1Objective..........................................................................................................................................15
2.2Parentfields.....................................................................................................................................15
2.3AddressingtheGap..........................................................................................................................21
2.4Theconceptofhabit........................................................................................................................21
2.5Perspectivesonhabit.......................................................................................................................26
2.6AttributesofHabit...........................................................................................................................31
2.6.1Habitsareacquisitions(learned)..............................................................................................31
2.6.2Habitsaretendencies(predictable)..........................................................................................32
2.6.3Habitsarepatterns...................................................................................................................33
2.6.4Habitsareextrapolators...........................................................................................................34
2.6.5Habitstendtoberewarding.....................................................................................................34
2.6.6Habitsarelatentuntilactivated...............................................................................................35
2.6.7Habitsareautomaticlike..........................................................................................................36
2.6.8Habitsareefficient....................................................................................................................37
2.6.9Habitsareshared(social)..........................................................................................................37
2.6.10Habitsareunique(individual).................................................................................................38
2.6.11Habitsareelastic(resilient)....................................................................................................39
2.6.12Habitsareplastic(malleable)..................................................................................................39
2.6.13Habitsareknowable(susceptibleofmetacognition).............................................................40
2.7Prototypedefinitionofhabits..........................................................................................................40
2.8ClassificationsofHabitinLiterature................................................................................................43
2.8.1Bylevelofintentionality(intentionalandunintentional)........................................................44
vi

 2.8.2Bymoralquality(goodandbad)...............................................................................................44
2.8.3Bylevelofvisibility(observableandhidden)...........................................................................44
2.8.4Bythelevelofcommonality(individualandcollective)...........................................................45
2.8.5Bythelevelofplasticity(rigidandflexible):.............................................................................45
2.9Adistinctionbetweenhabitvsinstinct............................................................................................45
2.10HabitandBehavior........................................................................................................................46
2.11Measurementofhabits..................................................................................................................49
2.11.1Measuringtheotherhabits:agapintheliterature.............................................................51
2.11.2MeasuresofMultiplePredeterminedHabits.........................................................................52
2.11.3Measuresforsingleandmultiplesemipredeterminedhabit................................................53
2.11.4Measuresforsingleandmultiplenonpredeterminedhabits................................................54
2.12Personenvironmentfit..................................................................................................................56
2.13Classificationsoffitinliterature....................................................................................................58
2.13.1Fitdefinedbypointofview(perceived,subjectiveorobjectivefit)......................................58
2.13.2Fitdefinedbylevelofspecificity(absoluteorrelativefit).....................................................58
2.13.3Fitdefinedbyuniontype(complementaryandsupplementaryfit)......................................59
2.13.4Fitdefinedbythelevelofbelonging(FitasaGestaltorasaprofiledeviation)....................59
2.13.5Fitdefinedbythelevelofobservability(physicalorcognitivefit).........................................59
2.14Theoreticalrelationshipsoffit.......................................................................................................60
2.14.1BehavioralIntentionandBehavior.........................................................................................60
2.14.2Otherrelationshipsoffit.........................................................................................................61
2.15Potentialadequacytomeasurehabits..........................................................................................61
2.16Potentiallimitationsofmeasuringhabitswithperceivedfit.........................................................63
2.17TechnologyFitandBehavior..........................................................................................................64
2.17.1TechnologytoPerformanceChain..........................................................................................65
2.17.2FitandAppropriationModel..................................................................................................66
2.18Definitionofhabittechnologyfit..................................................................................................67
2.19Technologyacceptance..................................................................................................................67
2.20Researchquestion1.......................................................................................................................69
2.21Hypothesis1and1a.......................................................................................................................70
2.22Theoreticalcontextforhabittechnologyfit:rationaleforitsselection.......................................72
2.23TheUnifiedTheoryofAcceptanceandUseofTechnology...........................................................73
2.23.1Performanceexpectancy........................................................................................................75
2.23.2Effortexpectancy....................................................................................................................76
2.23.3SocialInfluence.......................................................................................................................77
2.23.4Facilitatingconditions.............................................................................................................77
2.23.5InternalhypothesesoftheUnifiedTheoryofAcceptanceandUseofTechnology...............78
2.24Researchquestion2.......................................................................................................................82
2.25Hypotheses2and3........................................................................................................................82
2.26Hypotheses4and5........................................................................................................................84
2.27ResearchQuestion3......................................................................................................................84
2.28Conceptualframeworksummarized..............................................................................................85
2.29Researchmodel..............................................................................................................................86
2.30Summary........................................................................................................................................88
CHAPTER3METHOD...................................................................................................................................89
3.1Objective..........................................................................................................................................89
3.2ResearchProcedureOverview.........................................................................................................89
3.3Researchparadigm..........................................................................................................................90
3.4Researchmethodologyjustification................................................................................................91
3.5Researchdesign...............................................................................................................................93
vii

 3.6Datacollectiontechnique................................................................................................................93
3.6.1Datacollectionandtiming........................................................................................................95
3.7Samplingstrategy.............................................................................................................................96
3.7.1Unitofanalysis..........................................................................................................................98
3.7.2Samplesizeandresponserate.................................................................................................98
3.8Measurementofthevariables.........................................................................................................98
3.8.1Measurementprocedure..........................................................................................................99
3.9Measurementdevelopmentprocedureforhabittechnologyfit..................................................100
3.9.1Itemgeneration......................................................................................................................100
3.9.2Expertconsultation.................................................................................................................102
3.9.3QSortingExercise...................................................................................................................102
3.9.4Opensortexercise..................................................................................................................104
3.9.5Closedsortexercise................................................................................................................104
3.9.6AnalysesfortheQSortingexercises......................................................................................105
3.9.7ResultsfortheQSortingexercises.........................................................................................105
3.10Analysistechniques......................................................................................................................110
3.10.1Structuralequationmodelling..............................................................................................110
3.11Preteststudyprocedure.............................................................................................................112
3.11.1Questionnairerefinement....................................................................................................113
3.12PilotStudy....................................................................................................................................114
3.12.1Resultsofthepilotstudy......................................................................................................114
3.13Thefinalinstrument.....................................................................................................................116
3.14Mainstudy...................................................................................................................................117
3.15Dataanalysisprocedures.............................................................................................................117
3.15.1Datapreparation(Phase1)...................................................................................................119
3.15.2Reliabilitytest.......................................................................................................................119
3.15.3ExploratoryFactorAnalysis...................................................................................................120
3.15.4Confirmatoryfactoranalysis.................................................................................................121
3.15.5IndicatorsofModelFit..........................................................................................................122
3.15.6ValidityAssessment(phase3)..............................................................................................124
3.16Analysisprocessoverview:plannedvsreal.................................................................................125
3.16.1Analysissoftware..................................................................................................................128
3.16.2Issuesandconcerns..............................................................................................................129
3.17Summary......................................................................................................................................130
CHAPTER4ANALYSISANDRESULTS.........................................................................................................132
4.1Objective........................................................................................................................................132
4.2DescriptiveStatisticsoftheSample...............................................................................................133
4.3MeasurementReliability................................................................................................................134
4.4ExploratoryFactorAnalysis............................................................................................................135
4.5Confirmatoryfactoranalysis..........................................................................................................137
4.6MeasurementModelValidationBaseModel(UTAUT)...............................................................139
4.7MeasurementModelValidationExtendedModel......................................................................139
4.8MeasurementModelValidationModifiedModel......................................................................139
4.9ValidityAssessmentDiscriminantandConvergentValidity........................................................144
4.10TestHypotheses1and1a..........................................................................................................145
4.11TestHypothesis2......................................................................................................................147
4.12TestHypothesis3.......................................................................................................................150
4.13TestHypothesis4.......................................................................................................................153
4.14TestHypothesis5.......................................................................................................................154
4.15Redundancyanalysis....................................................................................................................155
viii

 4.16ModifiedModel...........................................................................................................................159
4.17TheoreticalCriteriaforPosthocModification............................................................................161
4.18ComparativeModelFit................................................................................................................163
4.19BriefReportofProblemsinAnalysis............................................................................................165
4.20Summary......................................................................................................................................165
CHAPTER5FINDINGSANDDISCUSSION...................................................................................................167
5.1Objective........................................................................................................................................167
5.2Positiverelationshipofhabittechnologyfitandbehavioralintention.........................................167
5.3 Moderation of age, experience and gender upon habit habittechnology fit and behavioral
intention...............................................................................................................................................169
5.4Basemodel.....................................................................................................................................170
5.4.1KeydeterminantsofintentioninUTAUT................................................................................170
5.4.2ModeratorsoftheUTAUTbasemodel...................................................................................171
5.4.3Basemodelfitwithdata.........................................................................................................173
5.5Extendedmodel.............................................................................................................................175
5.5.1Extendedmodelfitwithdata.................................................................................................176
5.6Modifiedmodelandotherfindings...............................................................................................177
5.6.1PosthocModifiedModel.......................................................................................................179
5.7Summary........................................................................................................................................181
CHAPTER6CONCLUSIONS........................................................................................................................182
6.1Objective........................................................................................................................................182
6.2Overviewofthesisobjectivesandresearchquestions..................................................................182
6.3Keytheoreticalcontributions........................................................................................................183
6.4Researchimplications....................................................................................................................184
6.5Practicalimplications.....................................................................................................................185
6.6LimitationsandFutureResearch...................................................................................................186
6.7Summary........................................................................................................................................189
REFERENCES..............................................................................................................................................190
APPENDIX1KEYCHARACTERISTICSOFHABITEXTRACTEDFROMDEFINITIONS..................................214
APPENDIX2SAMPLEOFDEFINITIONSBYDISCIPLINE...........................................................................218
APPENDIX3COREHABITDEFINITIONSGROUPED................................................................................219
APPENDIX4TYPEOFHABITMEASURED...............................................................................................223
APPENDIX5EMPIRICALEVIDENCEOFTHERELATIONSHIPSOFHABIT.................................................226
APPENDIX6QSORTINGEXERCISE:LISTOFITEMS...............................................................................229
APPENDIX7PARALLELANALYSIS...........................................................................................................231
APPENDIX8SPSSSYNTAX:RELIABILITY,EFA,ANDPARALLELANALYSIS...............................................232
APPENDIX9CORRELATIONMATRIX......................................................................................................235
APPENDIX10BASE,EXTENDEDANDMODIFIEDMODELSWITHSMARTPLS........................................236
APPENDIX11UTAUTWITHPLSGRAPH.................................................................................................238
APPENDIX12 ETHICSAPPROVAL...........................................................................................................240
APPENDIX13UNIONSANDINTERCEPTSCALCULATIONS.....................................................................241

ix

 LISTOFTABLES

Table2.1GapinLiterature.........................................................................................................20
Table2.2ObservancesofDefinitionsofHabitbyDisciplineGroup...........................................22
Table2.3KeyCharacteristicsofHabitExtractedfromDefinitions............................................23
Table2.4MapofPerspectivesofHabit......................................................................................27
Table2.5EmpiricalEvidenceofHabitUponBIandAB..............................................................47
Table2.6InternalHypothesesofUTAUT...................................................................................79
Table2.7EmpiricalTestsofUTAUTinPreviousResearch.........................................................81
Table3.1QSoringOpenExerciseSimilarityMatrix...............................................................106
Table3.2QSoringClosedExercisePopularPlacementMatrix............................................109
Table3.3MeasurementItemsandDemographicQuestions...................................................116
Table4.1DescriptiveStatisticsoftheSample..........................................................................134
Table4.2TableofReliabilities..................................................................................................135
Table4.3RotatedFactorMatrix..............................................................................................136
Table4.4CompositeandIndividualVariables.........................................................................137
Table4.5AssessmentofNormality..........................................................................................138
Table4.6MeasurementModelFitComparison........................................................................140
Table4.7CriteriaforConvergentandDiscriminantValidityCFA..........................................144
Table4.8ResultsHypothesis1.................................................................................................145
Table4.9ResultsHypothesis1a...............................................................................................146
Table4.10CriteriaValidationBaseModel(withoutModerators)........................................148
Table4.11CriteriaValidationBaseModel(withModerators)..............................................150
Table4.12CriteriaValidationExtendedModel....................................................................154
Table4.13CoefficientofdeterminationPLSvsSEM:BehavioralIntention............................156
Table4.14RedundancyAnalysis:R2andEffectSizeuponBehavioralIntention....................157
Table4.15CriteriaValidationModifiedModel(withModerators)......................................162
Table4.16StructuralModelFitComparison...........................................................................164

LISTOFFIGURES

Figure1.1ResearchModelSimplified........................................................................................11
Figure2.1RecurredBehavior.....................................................................................................19
Figure2.2HabitandDeliberation:DispositionalPerspective....................................................29
Figure2.3CoreTheoryofTechnologyAcceptance...................................................................68
Figure2.4UTAUTModel............................................................................................................75
Figure2.5ConceptualFramework..............................................................................................85
Figure2.6ResearchModel.........................................................................................................87
Figure3.1StagesintheResearchProcess..................................................................................90
Figure3.2Measurementdevelopmentprocedure..................................................................101
Figure3.3QSoringOpenExerciseDendrograms..................................................................107
Figure3.4AnalysisProcessDiagram........................................................................................118
Figure3.5PlannedvsRealAnalysisFlowchart.........................................................................127
Figure3.6ModeratorsinSEM..................................................................................................130
Figure4.1OrganizationoftheResults.....................................................................................132
Figure4.2ConfirmatoryFactorAnalysisBaseModel............................................................141
Figure4.3ConfirmatoryFactorAnalysisExtendedModel....................................................142
Figure4.4ConfirmatoryFactorAnalysisModifiedModel.....................................................143
Figure4.5BaseModel:UTAUT.................................................................................................147
Figure4.6ExtendedModel:UTAUT&HTF................................................................................153
Figure4.7VennDiagramofVarianceExplanationR2...............................................................159
Figure4.8NewSpecification:HabitTechnologyFitModel.....................................................161
Figure4.9PosthocModelModificationinProcess.................................................................162
Figure5.1RedundancyAnalysis:EffectSizeuponBehavioralIntention.................................178
Figure5.2NewModifiedModel:HabitTechnologyFitModel................................................180
Figure6.1HabitTechnologyFitModel....................................................................................184

xi

HABITSANDTECHNOLOGYFIT:

ASTUDYOFTECHNOLOGYACCEPTANCE

xii

 ABSTRACT

ThepurposeofthisresearchistostudyHabitTechnologyFit(HTF)asanapproachtocapturea
dynamicmixofhabitswhichareperceivedassalientfromtherespondentsperspective,and
theeffectofincludingitintheUnifiedTheoryofAcceptanceandUseofTechnology(UTAUT)
modelframework.TheHTFconstructanditsmeasurementweredevelopedinordertocapture
multiple nonpredetermined habits. All measures were semantically validated with QSorting
exercises, and their reliability was statistically confirmed. A crosssectional online survey was
distributedusingtheRespondentDrivenSamplingtechnique,reachingasampleof251adults
from 25 countries, who are SoftwareasaService users in public clouds and understand
English. Seven hedonic and utilitarian pieces of technology were included in the study. Data
was analyzed using covariance based Structural Equation Modeling and triangulated with
PartialLeastSquares.UnifiedTheoryofAcceptanceandUseofTechnology(UTAUT)wastested
alone,comparedwithanextensionofthemodelthatincludedHTF,andtoaposthocmodified
model.AllmodelswereassessedwithandwithouttheoriginalmoderatorsofUTAUT.Results
revealed a positive and significant relationship between Habittechnology fit and behavioral
intention,withastrongereffectinolderandmoreexperiencedindividuals.WhenUTAUTwas
extendedbyHTFthevarianceexplainedincreasedfrom52to58percent,andthemodelfitwas
slightly reduced. However, the probability for the original and extended models to fit other
samplesofthesamepopulationwaslessthan0.001.TheresultsalsoshowedthatHTFseffect
upon intention overlaps with the union of performance expectancy (PE) and effort
expectancys(EE)effect,butHTFprovidesanadditionalmarginofthevarianceexplained.This
suggested that UTAUT model could be simplified by using HTF in substitution of PE and EE
withoutmissingtheirexplanatorypower.TheposthocmodifiedmodelsubstitutedPEandEE
with HTF. It achieved the best model fit without moderators, and acceptable probability for
theorytests( =1.54,p=.10,RMSEA=.033andPCLOSE=.822). Despitethereductionof

variables, the posthoc model still explained 46 percent of the variance. Besides testing
previous theory, this research makes an original contribution to knowledge with the
conceptualizationanddevelopmentoftheHabitTechnologyFit,parsimoniousmodelthatwas
presented. These contributions addressed a gap in the literature of habit and technology
acceptance by incorporating multiple nonpredetermined habits. HTF measurement has the
xiii

limitations of selfreport instruments. Generalizability might be limited because conventional

probabilistic sampling techniques were not feasible for a ubiquitous population. Future
researchisneededtestingthemeasurementandposthocmodifiedmodelindifferentsettings,
inlongitudinaldesignsandusingdiversesamples;theHTFmodelisproposedasaparsimonious
alternativemodelforacceptanceoftechnology.

xiv

 CHAPTER1 INTRODUCTION
1.1 Objective

The aim of this chapter is to offer an introduction to the thesis. First, it presents the
foundation of a research strategy, research objectives and justification. Second, it
introducesthebackgroundsoftheresearchproblem,andpresentstheresearchquestions.
Third, it offers an overview of the methodology adopted, the key contributions to the
literature,andthelimitationsofthisresearch.Finally,itprovidesamapofthethesisandits
organization.

1.2 Researchobjectives

Thisthesishasthreeprimaryobjectives:

1. To study the relationship of the structure of habits and behavioral intention in

individuals,throughhabittechnologyfit.
2. To analyze the effect of including habittechnology fit, as a new construct, in the
UnifiedTheoryofAcceptanceandUseofTechnology(UTAUT)framework.
3. Toconductposthocmodelmodificationtoattempttoimprovetheresearchmodel.

Thefirstobjectiveisthemainfocusofthisthesisandguidesthisresearch.Itbringstogether
theory on habit, perceived fit and technology acceptance. The second objective compares
the seminal UTAUT framework with an extension that includes habittechnology fit. It
providesinsightaboutUTAUTatdifferentlevelsabout:themodelasawhole,itsvariables
and their relationships, and the effect of introducing habittechnology fit. The third
objectiveinvestigatesalternativemodelsfortechnologyacceptancethatmayincludehabit
technologyfit.

Other specific objectives are to conduct a literature review on habit, perceived fit and
technologyacceptance;conceptualizethehabittechnologyfitconstruct;designavalidand
reliablemeasurementscaleforhabittechnologyfittobeusedinthemeasurementmodel
of UTAUT; test the relationship of habittechnology fit and behavioral intention and its
moderators; test the original UTAUT model (as a whole model); test UTAUT extended by
1

HTF;determinetheuniquecontributionsofHTFintheframeworkofUTAUT;andconduct
posthocmodelmodificationtoprovideanimprovedmodel.

This thesis has collected data and additional demographic characteristics. Data was
collected from 25 countries. However, conducting group analysis or including other
demographicvariablesintheresearchmodelisnotpartofthescopeofthisthesis.

1.3 Background

Intheliteratureofhabit,habitshavebeenconsideredthemostsignificantdeterminantof
behavior(Chen&Chao2010;Chen&Lai2011;DeBruijn&Rhodes2010;DeBruijn&Van
Den Putte 2009; Gardner 2009; Liao, Palvia & Lin 2006; Loibl, Kraybill & DeMay 2011);
however,theyhavealsobeenoverlookedinmodernresearch.Oneofthereasonsforthe
paucityofhabitresearchfoundintheliteratureisrelatedtodoubtsonitsmeaningfulness.
Ajzen (1987) suggested that the relationship of pastbehaviorfuturebehavior is not
especially enlightening or insightful. Despite the general position of the psychology guild,
recent research such as that conducted by Limayem and Hirt (2003), Ouellette and Wood
(1998), Pahnila et al. (2011), Verplanken et al. (1998), Wood et al. (2002), among others,
held a positive point of view in regards to the meaningfulness of habit. Thus, habit
commencedtobeconsideredaproviderofuniqueinsightintothepredictionandcontrolof
behavior(Ouellette&Wood1998;Verplanken&Melkevik2008).Thismightexplainwhyan
ancientthemeisthatyounginthemodernresearchtradition.

However,Limayem,HirtandCheung(2007)pointedtotheimportanceofcapturingthereal
meaningofageneral,asopposedtoaspecific,habit(p.731)infutureresearch.Besides,it
remainsproblematicthatmeasurementofhabitandbehaviorareessentiallythesame,as
theformerissupposedtoexplainandpredictthelatter(Gardneretal.2011).Furthermore,
literaturesuggeststhathabitsaresystemsandstructuresthatinteractmutually,ratherthan
isolatedfrequentbehaviors(Bourdieu1984;Swartz2002;Wozniak2009).

In the literature of technology acceptance, several theories and models have been
proposed.SomeofthemostrelevanthavebeentheTheoryofReasonedAction(Fishbein&

Ajzen1975),theTechnologyAcceptanceModels(Davis1989;Venkatesh&Davis2000),the
Motivational Model (Davis, Bagozzi, & Warshaw, 1992), the Theory of Planned Behavior
(Ajzen1991),thecombinedTechnologyAcceptanceModelandTheoryofPlannedBehavior
(Ajzen1991;Taylor&Todd1995),theModelofPCUtilization(Thompson,Higgins&Howell
1991;Triandis1977),theDiffusionofInnovationTheory(Moore&Benbasat1991;Rogers
2003,originally1962),andtheSocialCognitiveTheory(Bandura1986;Compeau&Higgins
1995). As the models in the field grew in diversity, there was research which made
important attempts to synthesize attempts of synthesis. Thus, some of the most salient
theories and models were synthesized in the Unified Theory of Acceptance and Use of
Technology(UTAUT)(Venkateshetal.2003).

Four years after the publication of UTAUT, Venkatesh (Venkatesh, Davis & Morris 2007b),
one of the seminal authors of this theory, has described how technology acceptance
reachedastageofexpansioninwhichreplicationwithminortweaking(p.268)becamethe
pattern of recent years pattern of the last years. New research proposed models using
differentcombinationsofthesameelements,andminororiginality.Inthisway,Venkatesh
issued a challenge to find alternative theoretical perspectives. UTAUT was recently
extendedtoincorporatehabitasanimportantconstructintechnologyacceptance(Pahnila,
Siponen & Zheng 2011), and then the importance of habit was acknowledged in UTAUT2
whichadoptedthisvariableasanimportantcomponentofthemodel(Venkatesh,Thong&
Xu2012).

Thisthesiswasinitiallymotivatedbythepersonalexperienceofitsauthor.Asanenthusiast
of the possibilities that technology offers, he has embraced a substantial number of new
devices,operatingsystems,andsoftware.Awareoftheadvantagesofbeingaskillfuluserof
technology, he does not appreciate having to relearn how to use an ordinary computer
everynowandthen.Everycoupleofyears,eitheranewoperatingsystemoranofficesuite
comeswiththeallureofitsnewness.Butsignificantcostscomeassociatedwithadaptingto
the new in order to get back to the comforting stage of habitual utilization and minimal
thinking. These implicit costs impacting final users have been empirically studied for
radically new products. The outcomes showed that with greater innovation in the design,

usersstruggletofindaframeofreferencetounderstandradicalinnovation(GreganPaxton
etal.2002;Mugge&Dahl2013).Inthisway,manyproductsincludingtechnologiesmayfail
becausetheyaretoofarfromtheiruserreferences.

Literatureonpersonenvironmentfithasidentifiedperceivedfitasalatentvariablewhich
allows individuals complete cognitive manipulation of their evaluation. While responding,
individuals are allowed to define the salience of the various aspects of the variable to be
apprehendedintheiranswer(KristofBrown&Billsberry2012;KristofBrown,Zimmerman&
Johnson2005;Kristof1996).Thisconcepthasallowedassessingthecompatibilitybetween
people and organizations. However, extrapolation to habit (person) and technology
(environment)wasdeemedpertinentinthisthesis.

Gatheringtheelementsofthisbackground,theresearchproposalofthisthesiswasjustified
asfollows.

1.4 Justification

Firstly, it has been mentioned that habits might be the best predictors of behavior (Ajzen
1987;Ouellette&Wood1998).Secondly,literaturesuggeststhathabitscannotexistinpure
isolated forms (Bourdieu 1984; Swartz 2002; Wozniak 2009). Therefore, other habits may
also influence intention and behavior. However, the multiple habits contained in habitual
structures have been extensively overlooked in relationship with behavioral intention and
behavior(seeTable2.1).Thisthesisaddressesthisimportantgapintheliteratureofhabit.

The property of perceived fit measures may serve to capture salient habits, and salient
dimensions of habit from the respondent perspective (KristofBrown & Billsberry 2012;
KristofBrown,Zimmerman&Johnson2005;Kristof1996).Byapproachinghabitsinaway
thattakesadvantagefromthepropertiesofperceivedfit,thisthesismayprovideadditional
justification to conduct research. This is because its approach deals with multiple non
predeterminedhabits.Butitalsodealswiththeimpreciseandvariablenatureofaconcept
that belongs to the everyday languagehabit (Crossley 2013). Thus, this research is
justified as it addresses significant concerns in the literature of habit about the current

precision and completeness of definition and measurement for habits (Limayem, Hirt &
Cheung2007).

Theoretical frameworks such as UTAUT have recently been extended to incorporate the
important habit construct in technology acceptance (Pahnila, Siponen & Zheng 2011).
However,morethanasinglepredeterminedhabithasrarelybeenconsideredintechnology
acceptance research (see Table 2.1). Thus, addressing this gap in the literature of
technologyacceptancemayprovidejustificationfortheworkconductedinthisthesis.

Another aspect that justifies this research is that it would test UTAUT before adding
extensionsorperformingmodificationstothemodel.Testingwithaconfirmatoryanalysis
techniquewhichevaluatesthewholemodelsimultaneouslyisnecessarytoclaimtheoretical
suitability (Byrne 2010; Hair et al. 2010; Hair, Ringle & Sarstedt 2011). Since UTAUT was
createdwithexploratorytechniques,andisrarelytestedunmodifiedbyconfirmatorymeans
(SeeTable2.7),theconfirmatoryapproachofthisthesiscontributesinanaspectthathas
beenextensivelyoverlooked.

There are several reasons to have an interest in what it takes for people to accept
technology.Yet,thereasonsmayvaryfromtheperspectiveofthebeholder.Forcorporate
managers, employee acceptance of new systems might be important to support
performancegoalswithintheirorganization(Venkateshetal.2003;Westland&Clark2001),
toassurereturnoftheinvestmentandprofit,ortoproduceinternaldatathatcanbemined
toassistdecisionmaking(Huang,Wu&Chou2013;Morris&Venkatesh2010;Pan,Hackney
&Pan2008).Newproductdevelopersinthesoftwareindustrymayseekbusinessgrowthin
acceptance,whethertheydevelopadhocsoftwareorprovidesoftwareasaservicethrough
theInternet(Haldimann,Walter&Brenzikofer2014;Tyrvinen&Selin2011).Theinterest
to understand acceptance can also take the perspective of educators trying to teach
statistics to their students with didactic computer programs (Sydnor et al. 2014), of small
communities in need for a local surgeon being remotely assisted in delicate procedures
(Marescaux&Diana2014;Martini,Hewage&Nasralla2014).Despitetheperspectiveofthe
beholder, understanding technology acceptance has appeared as a subject of great

relevance in previous research. Thus, advancing toward better theory on acceptance may
continuetoprovideastrongrationaleforresearchinthisfield.

1.5 Significance

Empirical evidence of the relationship between habittechnology fit and behavioral

intentionwouldmeanthebetterthefitbetweenapersonshabitsandthetechnology,the
highertheintentionstoacceptandusetheproposedtechnology.Byprovidingsupportto
this relationship habittechnology fit may gain acceptance in the academic and
organizationalcommunity.

Whereassinglehabitstudiesmayexplaintowhatextentthehabitofusinganemailsystem
may impact on the utilization of the same email system, habittechnology fit may open a
new avenue of research. Encapsulated in habittechnology fit there could be a teenagers
habit of putting her mobile phone in her back pocket, influencing her intentions of
acceptanceofanicerbutwiderscreentablet.Habittechnologyfitmayalsoreflectthedeep
habitsinanaccountantsmind,affectinghisintentiontoacceptagreatonlinespreadsheet
that allows him to work simultaneously with other colleagues as he needs, but requires
learning too many new key shortcuts for the functions he already uses heuristically.
RegardlessofpreviousutilizationatechnologyX,theexistinghabitsmightbeinfavorofthat
technology or may oppose it. Such characteristic would be something quite unique about
habittechnologyfitthatotherapproacheshavenotprovided.

1.6 Researchquestions

Therewerethreequestionsdefinedforthisresearch:

RQ1Whatistheimpactofhabittechnologyfituponbehavioralintentioninthe
contextoftechnologyacceptance?(H1&H1a)

RQ2WhatistheeffectofincludinghabittechnologyfitintheUnifiedTheoryof
AcceptanceandUseofTechnologymodel?(H2&H3vsH4&H5)

RQ3Ifitcould,howcanhabittechnologyfitimprovethemodelspecification
of the Unified Theory of Acceptance and Use of Technology? (Posthoc
modification)
6

These research questions are linked to hypotheses which are developed in Chapter 2.
Researchquestion1referstotherelationshipbetweenhabittechnologyfitandbehavioral
intention.ThisquestionisaddressedbytestingHypotheses1and1a.Thefirstoneteststhe
direct relationship, and the second tests the relationship with threemoderators. Relevant
moderators define different strengths of the relationship at different levels of the
moderatingvariable.

Research question 2 refers to the effect of putting the measures of habittechnology fit
withinthemeasurementmodelofUTAUT,andthentestingthestructuralmodel.Inorderto
evaluatetheeffectofincludingthenewvariable,itisnecessarytotesttheoriginalUTAUT
model unmodified; then, include habittechnology fit and assess the differences between
models. Hypothesis 2 tests that all the internal relationships of UTAUT are as the theory
would expect. Hypothesis 3 tests the fit of the whole model with data. A good fit and
statistical significance would support the theory that posits the model as the best
theoretical structure. Hypotheses 4 and 5 repeat the same tests after including habit
technologyfit.Inthatway,itispossibletocomparewhichmodelhasbetterfitandbetter
statisticalsignificance.

Finally, research question 3 brings up the possibility of model improvement. The changes
havetoobeytheoreticalrelationships,butallowdifferentspecificationsofthemodel.Thus,
thisresearchquestionisaddressedbyconductingposthocmodelmodification.

1.7 ResearchMethodology

Thisthesishasdeemeditappropriatetofollowapostpositivistapproach,associatedwith
an objective approach to the study of social reality, which can only be imperfectly and
probabilistically apprehendable. The research on this thesis is framed in a quantitative
tradition, therefore in the deductive stream of research. A crosssectional design was
consideredappropriatetocollectquantitativeprimarydata,giventhetimelimitationofthe
researchproject.Selfadministeredquestionnairesareusedinseminaturalsettingswhere
respondents are asked to report. These decisions are based in assessing methodology

options from the literature of research methods, see (Blaikie 2010; Guba & Lincoln 1994;
Neuman2010).

Theprocessissummarizedinfivestages.Thefirststagebeganwiththeliteraturereviewof
habit, personenvironment fit and technology acceptance. It included identifying the
research problem, the main theoretical models; development of a conceptual framework,
research questions, and hypotheses. The second stage included measurement,
questionnaire and sample frame development. The scale development process included
itemgeneration,expertconsultationandacardsortingexerciseinordertoensurecontent
validity (DeVellis 2012; Moore & Benbasat 1991). A scale was developed for the HTF
constructonly.However,allUTAUTsmeasureswereexaminedandvalidated.Fourexperts
in areas of information systems, organizational communication, semiotics and linguistics
participated in the expert consultation, and 40 participants drawn from a public call in
Melbournetookpartinthecardsortingexercise.Then,apreteststudywasconductedto
refinetheinstrument,followedbyapilotstudy.Thethirdstageconsistedofthemainstudy
(onlinesurvey).Thefourthwasdataanalyses,andthefifthinterpretationandreporting.

Sevenhedonicandutilitarianpiecesoftechnologywereincludedinthestudy.Thesewere:
Facebook,GoogleDocs,MicrosoftOffice365,PayPal,Xbox360Onlinegaming,ZohoSuite,
Sales Force Cloud. Respondents selected the most familiar and the most unfamiliar
technology,andthesurveyfortheseoptionswasalternated.

The process of the analysis comprised (1) datapreparation, (2) reliability test,exploratory
factoranalysis(EFA)andconfirmatoryfactoranalysis(CFA),and(3)criterionrelatedvalidity
assessment, model testing and model modification. This evaluation required testing three
models and their internal hypotheses (original UTAUT, extended UTAUT and a modified
model).Thethreemodelsweretestedintwoversionsofthesamemodel(withandwithout
moderators).Thestructuralmodelsweretestedwithcovariancebasedstructuralequation
modeling (SEM) primarily, and crossvalidated with variancebased structural equation
modeling. Additionally, redundancy analysis and ftests were used to address collinearity
concerns.

1.8 Structureofthethesis

Chapteroneprovidesanoverviewofthethesis.Thischaptersetsresearchobjectivesand
justification.Itpresentstheresearchquestions,methodologyadopted,keycontributionsto
theliterature,andthelimitationsofthisresearch.Besides,itpresentsthestructureofthe
thesis.

Chapter two offers a review of the literature and theory of habit. This chapter presents
important concepts from the personenvironment fit to integrate the habittechnology fit
construct, identifies gaps in the literature, develops research questions and hypotheses,
developsaconceptualframework,andpresentstheresearchmodelofthisthesis.

Chapterthreediscussesthemethodologyapproachofthestudy,researchdesign,research
procedures, and scale development method. This chapter also introduces sample design,
anddataanalysisutilized.Throughoutthechaptertheadequacyoftheselectedprocedures
andtechniquesisdiscussed.

Chapterfourreportsadescriptivestatisticofthesample,detailstheresultsofreliabilityof
the scales, exploratory factor analysis, confirmatory factor analysis, and hypotheses test.
ChapterfouralsoshowsaVenndiagramthatillustratestheredundancyofeachvariables
effectuponbehavioralintention.

Chapterfivediscussesthefindingsofthisthesis.First,itpresentsthefindingsderivedfrom
Hypotheses1and1awhichrelatetotherelationshipofhabittechnologyfitandbehavioral
intentionanditsmoderators.Second,thefindingsderivedfromtestingHypotheses2and3
whichrefertotheoriginalUTAUTmodelarepresented.Third,thefindingsoftheextended
modelarediscussed,fromHypotheses4and5;andfourth,thechapterconcludeswiththe
discussionofthefindingsderivedfromdiscoveriesduringtheposthocmodelmodification.

Chaptersixconcludesthisthesis.Ithighlightsthetheoreticalcontributionsandimplications
ofthisresearch,acknowledgesthelimitationofthisstudyandsuggestsfutureresearch.

1.9 Keyconcepts

1.9.1 Habittechnologyfit

Habittechnology fit is a new construct developed in this thesis. Habittechnology fit was
defined as the degree to which an individual believes that using the technology is
compatible with his or her habits. The concept of perceived fit was borrowed from the
literature of personenvironment fit (KristofBrown & Billsberry 2012; KristofBrown,
Zimmerman & Johnson 2005) to shape the new construct. Habittechnology fit was
identifiedasapotentialdeterminantofbehavioralintention,moderatedbyage,experience
andgender.Thisthesisproposesintegrationwithinatechnologyacceptancemodel.

1.9.2 Behavioralintention

This concept refers to the strength of a persons intention to accomplish a behavior.

Behavioral Intention has been defined as an individual's positive or negative feelings
(evaluativeaffect)aboutperformingthetargetbehavior(Davis,Bagozzi&Warshaw1989,
p. 984; Fishbein & Ajzen 1975, p. 288). Behavioral intention is the main determinant of
actualbehavior,whichisameasureofthetargetbehavior(Davis1986).However,ithasalso
beenapproachedasalatentvariable(seeDavis1986;Liangetal.2010).

1.10 UTAUTtheoreticalframework

Habittechnology fit (HTF) was integrated in the framework of the Unified Theory of
Acceptance and Use of Technology (UTAUT) (Venkatesh et al. 2003). UTAUT posit
performance expectancy (PE), effort expectancy (EE), and social influence (SI) as
determinantsofbehavioralintention(BI).Theyalsoputforwardfacilitatingconditions(FC)
andbehavioralintentionasdeterminantsofactualbehavior(AB).Age,experience,gender
andvoluntarinessmoderatetheserelationshipsasbrieflyshowninFigure1.1.

10

 RESEARCHMODELSIMPLIFIED

Figure1.1ResearchModelSimplified (Source:Author)
(SeedetailedmodelinFigure2.6)

1.11 Keycontributions

Themainoriginalcontributionofthisresearchisthatitconceptualizeshabittechnologyfit
and empirically investigates its relationship to behavioral intention. In this theoretical
relationship,asystemofhabitsisconsidered,besidesthehabitwhichisexpressedbythe
same activities as the target behavior. Thus, the findings build upon previous research of
habit, and provide a significant contribution to the literature by addressing a gap where
multiplenonpredeterminedhabitshadnotbeenstudied.

Thecombinationofperceivedfitandhabits,asconceptualizedinthisthesis,anticipatesthat
thebetterthefitbetweenuserhabitsandagiventechnology,thehighertheintentionto
use it. This concept was empirically confirmed, which contributes to the theory of three
fieldshabit,personenvironmentfit,andtechnologyacceptance.Besides,thefindingson
hypothesized moderators, such as age and experience, confirmed the elastic property of
habits. This extends previous research on single habit and technology acceptance (see

11

Pahnila, Siponen & Zheng 2011; Venkatesh, Thong & Xu 2012). However, italso weakens
theoryongenderasamoderatorofthisrelationship.

This research contributes to theory by confirming some of the basic determinants

conceptualized in UTAUT, but weakening the applicability of its moderators and its
particular structure. It also weakens the applicability of the research model, consisting of
UTAUT plus habittechnology fit. By rejecting specific aspects of these models (base and
extended), the findings of this thesis may help to improve the generalizability of the
technologyacceptancemodels.

Animportantdiscoveryofthisresearchwastheredundancybetweenhabittechnologyfit,
performanceexpectancyandeffortexpectancy.Habittechnologyfitexplainsasmuchasthe
sumofperformanceexpectancyandeffortexpectancyplusanadditionalmargin.Thisstudy
foundthatbykeepinghabittechnologyfitinthemodelinsteadofperformanceexpectancy
andeffortexpectancy,thetotallossontheeffectsizeuponbehavioralintentionwouldbe
0.4%.Incontrast,bydroppinghabittechnologyfitauniquemarginof5.2%oftheeffectsize
would be lost. Furthermore, empirical evidence on semantic convergence and
differentiation led to a scenario where mixing these constructs was deemed inadequate.
Therefore, it might be difficult to specify habittechnology in the same model as
performance expectancy and effort expectancy. Such a discovery contributes to the
literaturerelatedtomeasurementofhabit,andperceivedcompatibility.Italsocontributes
to the literature of technology acceptance,where potentially valuable compatibilitybased
measures have been dropped without further assessment. These findings emphasize the
possibilityoferroneouslyassessing(errortypeII)discriminantvaliditywhenrelyingsolelyon
correlationbasedtechniques,suchasfactoranalysis.

Finally, one of the most important contributions to the theory of habit and technology
acceptancederivesfromtheposthocmodelmodification.Theonlymodelthatfoundmodel
fit and statistical significance for the model specification was a parsimonious one. Habit
technologyfitandsocialinfluencewerespecifiedasdeterminantsofbehavioralintention,
and behavioral intention as determinant of actual behavior. Referred to as the Habit

12

TechnologyFitmodel,thenewmodelrepresentsanoriginalcontributiontothetheoryof
habitandtechnologyacceptance.

1.12 Limitationsandfutureresearch

Oneofthelimitationsofthisresearchisthelevelatwhichitsfindingscanbegeneralized,as
this thesis used a nonprobability sample. Nonprobability samples are not ideal, yet
sometimes necessary when the elements of the population cannot be identified, and
therefore cannot be randomly selected (Blaikie 2010). However, this limitation was
addressed by using a RespondentDrivenSampling technique which reduces the bias by
homophily(Heckathorn2002).

Potentialselfreportbiasderivesfromaresearchchoiceofgettingdatafromaseminatural
setting,andmightrepresentalimitationofthisresearch.Whenpeopleareaskedtoreport
onthemselves,theremightbeagapbetweenthereportandreality(Blaikie2010).Thismay
alsoderivefromindividualsretrospectivelybecomingawareoftheirhabitsbythetracesof
theirunawareactions(Mittal1988),butonlybeingabletorecognizesomeoftheirhabits.
Thisresearchaddressedthislimitationbynotprovidingadditionalincentivesthatmayinvite
respondentstoreportdifferentlytoreality.However,thisselfreportremainsalimitationof
thisresearch.

TheHabitTechnologyFitmodelwasproposedasaresultofthisthesis.However,itneeds
validationinfutureresearch.Moreresearchisalsoneededtoestablishtheroleofculture
across several countries, and the use of the HabitTechnology Fit model in longitudinal
designs.

Facilitating conditions failed to achieve both convergent and discriminant validity in the
settingofthisresearch.Theconstructwasdropped,andthemodelhadtobetestedwithout
it. This thesis provided important results about the theoretical validity of UTAUT, but not
beingabletotestfacilitatingconditionsinthemodelisalimitationofthisresearch.Thus,
replicatingtheoriginalspecificationofUTAUTbyconfirmatoryanalysistechniquesremains
animportanttaskforfuturework.

13

1.13 Summary

This chapter offered an introduction to this thesis. It presented research objectives,

justification, research questions, an overview of the methodology adopted, the key
contributions to the literature, and the limitations of this research. It also presented the
organizationofthisthesis.Furthermore,thischapterpresentedtheneedtoinvestigatethe
relationship between habittechnology fit and behavioral intention in the framework of
UTAUT.

14

 CHAPTER2 LITERATUREREVIEW
2.1 Objective

Theobjectiveofthischapteristoprovideasystematicreviewandanalysisoftheliterature
ofhabitandbehavior.Thisreviewaimstoidentifyresearchgapsintheliterature,andthe
justification of addressing them. This chapter also explores the concept of fit in order to
provide an alternative operationalization for habits, and theory on technology acceptance
withthepurposeofcontextualizingtherelationshipsofhabitsandbehavior.

Thischapterconcludeswiththeconceptualizationofanewconstruct,habittechnologyfit.
Basedonthereviewoftheoryandempiricalevidence,relationshipsforthenewconstruct
arehypothesizedinthecontextoftheUnifiedTheoryofAcceptanceandUseofTechnology.

2.2 Parentfields

Habits are considered a major determinant of behavior; however they have been
overlookedinpastresearch.Thepsychologyguildrecognizesthathabitstrengthasthebest
predictor of future behavior (Ajzen 1987). Paradoxically, habit does not appear as an
important construct in most contemporary human behavior research models as Ouellette
andWood(1998)describe.

This can be difficult to believe because on one hand, the term habit may seem too
quotidian, and it might be easy assuming significant presence in the peer reviewed
literature. Besides, the proliferation of selfhelp literature describing the importance of
habitsinpersonalperformancecanmisleadtheperceptionaboutthematurityoftheoryon
thefield.

The reason for the exclusion of habit might be related to the doubts about the
meaningfulnessofitsroleinlinkingpastwithfuturebehavior.OuelletteandWood(1998)
explain how most graduate psychologists learn a generally accepted dichotomy of habit.
Habitsasfrequencyofpastbehaviorarearobustpredictoroffuturebehavioraccordingto
Triandis(1977),Thompson,HigginsandHowell(1991),Bergeronetal.(1995),andDennis,
WixomandVandenberg(2001).Butthisisimmediatelycontrastedwithargumentsagainst
15

the significance and value of Triandis statement. The relation of pastbehaviorfuture

behaviorisnotespeciallyenlighteningorinsightful.Forinstance,sayingthatapersonison
timeforherappointmentsbecauseshehasatendencytobepunctual,doesnotprovideany
newinformationorunderstanding(Ajzen1987).

Theconceptofhabitevolvedfromsimplerepetition,experienceorpastbehaviortomore
robustconceptualizations.Despitethegeneralpositionofthepsychologyguild,duringthe
past three decades research on habit reemerged. A number of researchers such as
LimayemandHirt(2003),OuelletteandWood(1998),Pahnilaetal.(2011),Verplankenetal.
(1998),Woodetal.(2002)supportedthemeaningfulnessofhabitasaproviderofunique
insightintothepredictionandcontrolofbehavior(Ouellette&Wood1998).

InInformationSystems(IS)literature,extendingmodelsbasedonpersonalcharacteristics
suchashabitisencouragedasamatterforfutureresearch(Baloh2007;Junglas,Abraham
& Ives 2009). Maruping & Agarwal (2004) researchers are encouraged to extend their
theoriesintotheintersectionoftechnology,humans,andtasks(p.15).

In this regard, habit may well be a suitable representative of personal characteristics.

Abelson(Abelson1981;Verplanken,Myrbakk&Rudi2005)suggestedhabitsareascriptofa
personal kind, Lankton et al. (Lankton, Wilson & Mao 2010) recognizes habits as self
identifying, and Norman (Norman 2011) makes reference to habits reflecting a persons
senseofidentity.Therefore,byresearchingonhabitsthisthesismightbeaddressingsuch
calls.

Besides the call for personal characteristics, habit has received direct calls for future
researchinInformationSystems(IS)andparticularlyinthetechnologyacceptancemodels.
Given the importance of habit in web site usage, and by extension in the context of
informationsystemswheremuchuncertaintyispresent,itseemsnecessaryforresearchers
to seriously examine the role of habit in technology adoption (Liao, Palvia & Lin 2006, p.
481).Habitisamajorfactorbehindbehavioralintentionsandshouldbeincludedinmodels
suchasTAM(Gefen2003,p.10).ThedevelopersoftheUnifiedTheoryofAcceptanceand
UseofTechnology(UTAUT)haverecommendedextendingUTAUTbyintegratinghabitinto

16

it as a path of research in order to increase the explanatory value of UTAUT (Pahnila,

Siponen&Zheng2011;Venkateshetal.2003).Sofar,thiscallhasbeenaddressed,butonly
single predetermined habits have received attention (see Section 2.11.1 Measuring the
otherhabits:agapintheliterature).

Dennis et al. (2001) proposed a fit and appropriation model and theoretically identified
habitsasadeterminantofappropriationofgroupsupportsystems;howeverhabitswere
excludedfromtheempiricalstudy.Pahnilaetal.(2011)effectivelyintegratehabitwithinthe
frame of UTAUT for the Chinese eBay. When Pahnila et al. published their 2011 article,
consistent with this literature review, no other study had extended UTAUT with the
constructofhabit.Thefindingssupportedapositiveimpactofhabitsuponactualbehavior,
measuredwiththeSelfreporthabitIndex(Verplanken&Orbell2003).

FinallyVenkateshetal.(Venkatesh,Thong&Xu2012)publishedasecondversionofUTAUT.
UTAUT2extendedUTAUTwithhedonicmotivation,pricevalue,andhabit.Habitwastested
inbehavioralintentionandactualbehaviorandbothrelationsresultedsignificant.

Inregardstohabit,Pahnila,SiponenandZhengs(2011)extensiontoUTAUTandVenkatesh,
ThongandXus(2012)UTAUT2mayhavecoveredanimportantgapintheliterature,which
makesthemrelevantforthisthesis,althoughnomorethanotherstudiespresentedinTable
2.1. Other approaches to measuring habits might significantly expand these contributions
on habits in technology acceptance. Limayem, Hirt and Cheung (2007) affirmed that an
important task for future research would be capturing the real meaning of a general, as
opposedtoaspecific,habit(p.731).BesidesGardneretal.(2011)andAjzen(2002)would
agreeinthatamoreappropriateapproachwouldrelyonanoperationalizationofhabitthat
isindependentofthebehavioritissupposedtoexplainandpredict(p.14).

Figure 2.1 shows three examples in which single predetermined habit has been tested
empirically(seealsoTable2.5EmpiricalEvidenceofHabitUponBIandABorAppendix5).
In UATUT2, the relationship of habit has been expressed as in the Example 1 of the fore
mentionedfigure(Venkateshetal.2003).

17

In the three examples, the researcher knows and wants to measure one habit and one
behavior.Letterh1representsonehabitmeasuredatthetime,andb1representsabehavior
beingmeasured.Thesubscript1identifiesthespecificbehavior.Forexample,ifb1isthe
behavior of using a PC, h1 is the habit of using a PC. The examples in Figure 2.1
operationalize habit in a way that hardly can be considered independent of the behavior,
simply because the target behavior is the same as the habituated behavior. In the next
section, Table 2.1 (Gap in Literature) shows that most of the research published so far
concentratesonsinglepredeterminedhabit,whichusuallytranslatestohabitwithimpact
uponarecurredbehavior.

Agapintheliterature,wheresignificantcontributionscanstillbemade,mayderivefrom
thetheorywhichsuggeststhathabitsdonotexistinapureandisolatedform.InSection2.7
(Prototypedefinitionofhabits),thisthesisdevelopedanapproximationtothedefinitionof
habit.ConsistentwithWozniak(2009),Swartz(2002),andBourdieu(1984)habitsarecalled
systems and structures, which indicates relationships of interaction and support among
habits.However,itisrarefindingdirectreferencestotheinteractionbetweenhabitsinthe
literature of habit and behavior, therefore empirical evidence is also limited (see Section
2.11.1Measuringtheotherhabits:agapintheliterature).

18

RECURREDBEHAVIOR
Example1

Example2

Example3

Figure2.1RecurredBehavior (Source:Author)

Basedonthepreviousarguments,thisthesishasthesupporttoproposethatincludingthe
otherhabitsinempiricalresearchwouldconstituteanoriginalandsignificantcontribution
totheliteratureofhabit,informationsystemsandbehavior.

Table 2.1 classifies the different measures of habit in the literature. The classification
considershowmanyhabitsarebeingmeasuredatthesametime,andtowhatextendthe
habitsbeingmeasuredarepredeterminedbytheresearcher.Thistableillustratesthatmost
of the attention has been placed on measuring single predetermined habits, and multiple
nonpredetermined habits have been overlooked in the literature. Section 2.11.1
Measuringtheotherhabits:agapintheliterature,providesmoredetailsonthecriteriaof
classification.

19

 GAPINLITERATURE
Single Multiple
SP MP
Predetermined 51 4
(83.6%) (8.2%)
SS MS
Semi
1 3
predetermined
(1.6%) (4.9%)
SN MN
Non
1 0
predetermined

(1.6%) (0.0%)
TotalArticles Measurementof Reported
Singlepredeterminedhabit(SP) 51 83.6%
Multiplepredeterminedhabits(MP) 4 8.2%
Singlesemipredetermined(SS) 1 1.6%
58 Multiplesemipredetermined(MS) 3 4.9%
Singlenonpredeterminedhabit(SN) 1 1.6%
Multiplenonpredeterminedhabits(MN) 0 0.0%
TOTALCASES 61 100.0%

1(EscobarRodrguez&CarvajalTrujillo2013) 21(Limayem&Cheung2008) SP 41(Trafimow2000)SP

SP
2(Han&Farn2013)SP
3(Huang,Wu&Chou2013)SP
4(Kangetal.2013)SP
5(Klckner2013)SP
6(Nikou&Bouwman2013)SP
7(Raman&Don2013)SP
8(Tseng,Chang&Woo2013)SP
9(Venkatesh,Thong&Xu2012)SP
10(Barnes2011)SP
11(Chen&Lai2011)MP
12(Loibl,Kraybill&DeMay2011)SP
13(Norman2011)SP
14(Pahnila,Siponen&Zheng2011)SP
15(DeBruijn&Rhodes2010)SP
16(Guetal.2010)SP
17(Lankton,Wilson&Mao2010)SP
18(deBruijnetal.2009)SP
19(DeBruijn&VanDenPutte2009)SP
20(Gardner2009)SP
Table2.1GapinLiterature
22(Wu&Kuo2008)SP
23(DeBruijnetal.2007)SP
24(Limayem,Hirt&Cheung2007)SP
25(Liao,Palvia&Lin2006)SP
26(Thgersen2006)SP
27(vanEmpelen&Kok2006)SP
28(Verplanken2006)SP
29(Honkanen,Olsen&Verplanken2005)SP
30(Kim&Malhotra2005)SP
31(Wood,Tam&Witt2005)SP
32(Gefen2003)SP
33(Klckner,Matthies&Hunecke2003)SPS
SMPSSMS
34(Limayem&Hirt2003)SP
35(Limayem,Cheung&Chan2003)SP
36(Limayem,Hirt&Cheung2003)SP
37(Verplanken&Orbell2003)SPSN
38(Bamberg&Schmidt2003)MS
39(Orbelletal.2001)SP
40(Saba,Vassallo&Turrini2000)SP
42(Verplanken&Faes1999)MP
43(Ouellette&Wood1998)SP
44(Saba&DiNatale1998a)SP
45(Saba&DiNatale1998b)SP
46(Sabaetal.1998)SP
47(Verplanken,Basetal.1998)MS
48(Bergeronetal.1995)SP
49(Verplanken1994)MP
50(Towler&Shepherd1992)SP
51(Ajzen1991)N/A
52(Montano&Taplin1991)SP
53(Bagozzi&Warshaw1990)SP
54(Charng,Piliavin&Callero1988)SP
55(Mittal1988)SP
56(Wittenbraker,Gibbs&Kahle1983)SP
57(Bagozzi1981)SP
58 (Landis, Triandis & Adamopoulos 1978)
MP
(Source:Author)

20

2.3 AddressingtheGap

Based on the principles of fit previously explained in Sections 2.12 to 2.16 , this thesis
argues that borrowing personenvironment [PE] theory and perceived fit measurement
techniques, may allow a combined operationalization with habits. While responding to an
integrated measure of fit and habits, individuals would be able to select any habit that
comes to their mind. They could select any characteristic that increases or decreases the
compatibility between the person and the technology. Furthermore, supplementary and
complementary compatibility between habit and technology could be captured in a single
measure(seeSection2.13.3).

Theproductofextrapolatingperceivedfittothestudyofhabitsintechnologyacceptance
derivesinanewconstructdevelopedinthisthesis:habittechnologyfit.Therelationshipsof
habittechnologyfit,behavioralintentionandactualbehaviorcanbecontextualizedinthe
frameoftheUnifiedTheoriesofAcceptanceandUseofTechnologyUTAUT1&2(Venkatesh
etal.2003;Venkatesh,Thong&Xu2012).

2.4 Theconceptofhabit

This section presents one issue in the literature of habits, this is, the imprecision of the
conceptitself.Asmentionedintheintroductionofchapter2,futureresearchcontinuesto
callforbetterdefinitionsofhabitthatmayleadtobetteroperationalization(Limayem,Hirt
& Cheung 2007). However, the main problem continues. The concept belongs to the
everydaylanguage,whereitsdefinitionisvariableandimprecise(Crossley2013).

Thus, section 2.5 is dedicated to the diverse perspectives of habit. Then, section 2.6 will
revise literature that describes the attributes and characteristics of habits. These two
sections show that the short dictionarylike approaches to define habit tend to be
reductionist by oversimplifying the essence of the concept. Therefore we commence this
review with a picture of the definitions of habit adopted or generated within diverse
standingpoints.

21

Commencingwithasimplecount,thenumberofdefinitionsofhabitsfoundperdiscipline
might suggest the attention each field of knowledge has paid to the topic. Thus, a search
(habit isOR habitsare) wasconducted in the 43 databases of ProQuest Central. 5530
results were revised looking for definitions, and 98 Peerreviewed articles containing 135
definitionsemerged.Resultswereclassifiedinmoredisciplinegroups,andobservancesby
discipline appeared as in the following proportions: 26% of the definitions of habit were
found in Medicine and Health, 18% in Humanities, 17% Business and Economics, 16%
Psychology,11%inEducation,6%inScienceandTechnology,and6%inSocialSciences(See
Table2.2).

OBSERVANCESOFDEFINITIONSOFHABITBYDISCIPLINEGROUP
Discipline Frequency Percent
Healthandmedical 35 26%
Humanities 24 18%
Businessandeconomics 23 17%
Psychology 22 16%
Education 15 11%
ScienceandTechnology 8 6%
SocialSciences 8 6%
TOTAL 135 100%
Table2.2ObservancesofDefinitionsofHabitbyDisciplineGroup (Source:author)

Withinthedefinitionsofhabit,itwasfoundthatwordssemanticallygroupedasbehavioral
tendency (for example disposition, inclination, predisposition, proclivity, propensity is an
essential characteristic of habits (see Appendix 4 and Appendix 3 for full details on how
groupswereformed).Thischaracteristicwasthegreatestcommonalityamongthesefields
of knowledge (it appeared in 30% of the definitions). The second most commonly found
characteristic(observedin11%ofthedefinitions)conceptualizedhabitasapattern.The
third most frequent reference to habit was the word behavior (seen in 10% of the
definitions). After that, other distinctive characteristics of habits appeared with some
frequencyacrossdisciplines,suchasbeingways(7%),acquisitions(4%),responses(4%)
and routines (4%), among many others (see Table 2.3). Fortynine different words were
usedaskeycharacteristicsofhabitbeforegroupingthem.

22

 KEYCHARACTERISTICSOFHABITEXTRACTEDFROMDEFINITIONS
KeyCharacteristic Frequency Percent
Tendency 41 30%
Pattern 15 11%
Behavior 14 10%
Ways 9 7%
Acquisition 5 4%
Response 5 4%
Routines 5 4%
(Other) 41 30%
TOTAL 135 100%
Table2.3KeyCharacteristicsofHabitExtractedfromDefinitions (Source:Author)
(SeeAppendix1)KEYCHARACTERISTICSOFHABITEXTRACTEDFROM
DEFINITIONS

However,itcallstheattentionthatwordsusedindefinitionsasfundamentalcharacteristics
of habit can be as diverse as they are. There are as many observances of definitions
containingwordssemanticallygroupedastendency,astherearewordsthatcouldnotbe
groupedandwordsappearingwithafrequencylowerthan5inasampleof135.Giventhe
complex nature of habits, it is not surprising that in research there has been difficulty to
definetheconceptandeveninattemptingtomeasureit.

AccordingtoVerplankenandMelkevik(2008)oneofthemostlyignoreddiscussionsarethe
ones concerning the conceptualization of habit. It is still vague what it can be considered
habitual.Productofsituation,appropriateconceptualizationandoperationalizationarestill
amatterofdebate(Lallyetal.2010).

According to the content observed in the definitions of habit, two identifiable trends
appearedinoursample.OneseemstobealignedwithJohnDeweysdefinition(originally
published in 1922) (Dewey 2002), the second apparently to William James (James 1890).
Dewey presents elements of predisposition to respond as a characteristic pattern of a
person. Dewey also suggests that tendency to repeatacts are not the essence ofhabit as
theirtendencytorepeatwouldbeanincidentofmanyhabitsbutnotofall(Dewey2002,
p.42).

23

Thesecondtrendobservedpaysmoreattentiontosequencestriggeredbycontextualcues;
acts are considered automatic responses or an outcome of a tendency to repeat past
behavior.Inotherwords,thissecondstreamfocusesmoreonmechanicallikeautomaticity.

For Dewey The essence of habit is an acquired predisposition to ways or modes of

response (Dewey 2002, p. 42). Within the sample of definitions in this review, 25% have
takenthedefinitionorverydistinctiveelementsfromDewey(Abowitz2011;Berk&Galvan
2009; Biesta 2007; Brinkmann 2007; Brockelman 2002; Charmaz 2002; Crissman 1942;
Cutchin 2000, 2007; Garrison 2002; Guerreiro, Pereira & Frezatti 2006; Hedoin 2009;
Hodgson2009;Hodgson2010;Kemp1998;MacMullan2005;Miller2010;Nakamura2009;
Ralston 2011; Reynolds 1981; Ronald Lee 1998; Stengel 2010). This element is present in
every field, except for oneScience and Technology, where Information Systems was
included.

Ontheotherhand,James(1890)defineshabitassequencesofbehaviorthathavebecome
virtually automatic (p. 107). Of the definitions in the researchers sample, 13 definitions
(9.6%) contain the essential characteristics of this definition (Archer 2010; Bansal 2011;
Bonneetal.2007;Brder&Schiffer2006;Canin,Dolcini&Adler1999;deNooijer,Onnink&
vanAssema2010;Fujii&Kitamura2003;Lawrence,Evans&Lees2003;Mair&Berginseers
2010;Mooreetal.2006;Schmuck&Vlek2003;Theuvsen2004;Thgersen&Mller2008;
Yoon 2011). This definition was observed in every discipline of the sample of definitions,
exceptintwofieldsEducationandHumanities(seeAppendix2andAppendix4).

Definitionsofhabitinsocialsciencesarescarce.Thereareatleasttworeasonsforsucha
minor appearance of definitions in the field. Firstly, concept was removed from the
vocabularyinsociologyinoppositiontothebehavioristmechanismsthatdenytheagencyof
the individual. The exile of the term is particularly notorious between 1940 and 1970.
Secondly, Pierre Bourdieu came up with a sociologicalacceptable alternative to habit
habitus(Crossley2013).

Theanalysisof135definitionsofhabitfoundin98peerreviewarticlesledtheresearcherto
synthesize key commonalities. These commonalities areremarkably similar and consistent

24

to Wacquants (2005) synthesis of Bourdieus concept of habitus: lasting dispositions

[tendency]ortrainedcapacities[acquisition]andstructuredpropensities[pattern],andto
think,feel,andact[behavior]indeterminateways[](Collet2009,p.421).

Crossley (2013) explains that throughout Bourdieus publications, there is no single

authoritativedefinitionofhabitus.Bourdieurisesandexplainshisviewsonhabituswithout
reducingittoashortdefinition.Ononehand,thatprovidesarichviewoftheconcept.On
theotherhand,itmakesitdifficulttoapproachhabitusasameasurablevariable.But,that
is not unanticipated as Bourdieu had a personal position against the positivist approach
(Collet2009).

Section 2.5 will discuss the trends and perspectives around habit, and particularly the
dichotomy of empiricism and transcendentalism. However, the synthesis of definitions of
thisthesisdoesnotreflectthelevelofcentralityoftheautomaticeffectwhichisdominant
inthetrendafterJames(1890).ThiscentralityisleadingtheInformationSystems(IS)field,
whichmighthaveinheritedsomeofitsperspectivesfrom(James1890).

KeycharacteristicsofhabitarediscussedinCrossley(2013),Collet(2009),Hodgson(2010),
Klckner et al. (2003), Lindbladh and Lyttkens (2002), Pahnila et al. (2011), and Polites
(2009),butarefrequentlyexcludedfromthecoredefinitions.Crossley(2013)discusseshow
the concept of habit is particularly complicated as it belongs to the everyday language,
whereitsdefinitionisvariableandimprecise.Thus,theintrinsiccomplexityofhabitleadsto
aneedformorecomprehensiveconceptualizations.

Thissectiondiscussesacontentanalysisappliedtoasampleof135definitions.Tendency,
patternandbehaviorwerethemostfrequentlyfoundkeytermsusedtodefinehabit.These
terms account for 52% of the cumulative percentage of the observances of key defining
terms. There are a number of fundamental characteristics expressed in the synthetic
definitionsofeverydiscipline.However,thesedefinitionsdonotcaptureotherfundamental
characteristicsofhabit.

25

Section 2.6 presents thirteen attributes explaining habits. These were extracted from a
deeperanalysisofthepeerreviewedliteratureonhabitsandbehaviorusingNVivov.10.The
themeswereusedtosuggestessentialattributesandclassesofhabit(Section2.6)froma
holisticperspective.However,tounderstandthisperspectiveitbecomesnecessarytodraft
amapofthedifferentperspectivesonhabit.

2.5 Perspectivesonhabit

The first stream of contemporary thought identified in literature wasbehaviorism (Becker

1992; Hull 1943; Watson 1998), which is a school of psychology based on the stimulus
response perspective. Their exponents suggest that habits are always based on rational
choice,andeverypieceofinformationistakenintoaccountfordecisionmakingwhileacting
by habitual behavior. This view shares the mechanist concept of habit with the Biological
Perspective (James 1890; Watson 2008) and the economics mainstream in the notion of
habitasasequentiallycorrelatedbehavior(Becker1992)wherethepastcorrelateswiththe
future behavior (Hodgson 2010; Polites 2009). Foucault (Foucault 1973) refers to these
perspectivesasanempiricistpointofview,characterizedbyitsscientificconceptionofthe
humanbeing.

Table 2.4 shows a map of some of the most outstanding perspectives identified in the
literature of habit. Although these perspectives can be described with different names by
differentauthors,theytendtofallintotwomaincategories:empiricalandtranscendental
perspectives as described in (Foucault 1973). The empirical perspective takes a scientific
approach to understand human beings, whereas the transcendental takes a philosophical
approach.TostressthisperspectivedistinctionBourdieusetasidethewordhabitandused
theLatintermhabitus(Crossley2013).

26

 MAPOFPERSPECTIVESOFHABIT
Author BEH BIO CMP DIS ECO EMP SCB SOC SSD TRSC
(Aarts,H.&Dijksterhuis,AP2000) 7
(Abelson1981) 4
(Aristotle1976) 1
(Bartlett1997) 4
(Beck2004) 5
(Becker1992) 3 5
(Bourdieu1990) 2,5 1
(Bover1991) 5
(Dewey2002,reprint1922) 3 1
(Hodgson2010) 3 5
(Hull1943) 7
(Hume1984) 1
(Husserl1970) 1
(James1890) 4 3
(Kant2007)
(Lindbladh&Lyttkens2002) 5
(Mauss1979) 1
(MerleauPonty1965) 1
(Pavlov1911) 1
(Sheeranetal.2005) 7
(Triandis1979) 4
(Veblen1898) 3
(Verplanken 2006) (Verplanken & Orbell
7
2003)
(Watson1998) 3 4
(Weber2004) 1
(Wood & Neal 2009) (Melcalfe & Mischel
6
1999)
Simon(2001;1974) 2
BEHBEHAVIORISM
CMPCOGNITIVEMOTIVATIONALPERSPECTIVE
1(Crossley2013)
DISDISPOSITIONALPERSPECTIVE
2(Collet2009)
SCBSOCIALCOGNITIVEBEHAVIORISTSYNTHESIS
3(Hodgson2010)
BIOBIOLOGICALPERSPECTIVE
4(Klckner,Matthies&Hunecke2003)
SSDSCHEMAANDSCRIPTDISCUSSION
5(Lindbladh&Lyttkens2002)
SOCSOCIOLOGICALPARADIGM
6(Pahnila,Siponen&Zheng2011)
ECOECONOMICSPARADIGM
7(Polites2009)
EMPEMPIRICALPERSPECTIVE(SCIENTIFIC)
TRSCTRANSCENDENTALPERSPECTIVE(PHILOFOPHICAL)
Table2.4showswhichauthorswereidentifiedintheliteratureasrepresentativeofatheoreticalperspective.
Each column represents one theoretical perspective. The headings of the columns are the abbreviations of a
perspective.Decodedabbreviationsarefoundabovethisparagraphattheleftofthetable.Thenumbers17
representsevensourcesthatidentifyagivenauthorwithatheoreticalperspective.Thesenumbersareplaced
attheintersectionofrowsandcolumnstoconnectoneauthortooneperspective.
Table2.4MapofPerspectivesofHabit (Source:Author)
27

Most perspectives today oppose behaviorism or pure behaviorism. For example, the
CognitiveMotivational Perspective (Polites 2009), the Dispositional Perspective (Hodgson
2010),andtheSocialCognitiveBehavioristSynthesis(Wood&Neal2009).

The CognitiveMotivational Perspective (CMP) focuses on the individuals goals during the
development of habits. Whereas behaviorism would see a direct connection between
stimuliandresponse,CMPconsidersgoalsasmediators.CMPseespositivereinforcement
as the strengthener of the link between goal and action. In a way that when a similar
situation emerges, the representation of the goal is automatically activated and the
behavior occurs without the need of conscious thought. Although CMP uses some of the
lexiconofbehaviorism,suchaspositivereinforcementandautomaticity;itistheindividuals
desiredeffectthatdrivestheperformanceofbehavior(Polites2009;Sheeranetal.2005).

Thedispositionalperspective(DIS)seeshabitsasacquireddispositions.DISliteratureposits
that choice, but also instincts and other nondeliberative channels, inform habits. DIS
disagreeswithbehaviorisminthatchoiceistheoriginofhabitandindividualscanuseallthe
pieces of information to make deliberated choices. DIS relies on evidence that acts of
deliberationareledunconsciousbrainprocessesindicatingthatindividualsaredisposedto
their choices before they are aware of their decision. Habit efficiently encapsulates past
adaptive behavior (Hodgson 2010). According to this perspective choice is real but it is
caused, it is a contingent outcome of habits. Figure 2.2 shows the contrast between the
behavioristperspectiveandDIS.

28

HABITANDDELIBERATION:DISPOSITIONALPERSPECTIVE
BehavioristPerspective

DispositionalPerspective

Figure2.2HabitandDeliberation:Dispositional (Hodgson2010)
Perspective

SocialCognitiveBehaviorist Synthesis (SCB) perspective focuses on the processes of habit

chance.SCBpositsthatpeoplecaninhibithabitsbyexcretingcontrolafterahabitcuehas
activated the response in the individuals memory. In this way, habits can be broken.
However,habitisconceivedasatypeofautomaticitywithcharacteristiccueingofbehavior
that does not depend on goals or intentions, but cognitive associations. It differs from
behaviorisminthatitacknowledgeshumanactionsaspurposive(Pahnila,Siponen&Zheng
2011; Wood & Neal 2009). In the aspect of goals, SCB differs from the Cognitive
MotivationalPerspective(CMP)mentionedbefore.

Theschemaandscriptperspectiverelyonthenotionofschematicorheuristicdecisionsor
cognitive mechanisms which are deployed in automatic alike ways. Similar to the
DispositionalPerspective(DIS)itrecognizescognitivelimitations,andseeshabitasaformof

29

saving those resources. However, this perspective considers that habits understood as
schemas or scripts are not accessible through selfreports due to their unconscious
character.Thisbeliefledtoameasurementwhichisnotdependentfromselfreports,the
Response Frequency Measure or RFM (explained in Section 2.11 Measurement of habits).
Triandis model (Triandis 1979) was one of the first incorporating this concept. Klckner,
MatthiesandHunecke(2003)identifytheschemaandscriptnotionasconflictingwiththe
mechaniststrictviewofthebiologicalandbehavioristperspective.

In the sociological perspective (SOC) the individual surpasses biological boundaries to act
free.Thephilosophicalfoundationoffreewillandvolitionoftheindividualoragentledto
theadoptionofaslightlydifferenttermtorefertohabitshabitus.Thisishabitus,anovert
rejection to behaviorism where habit denotes mechanical behavior in response to a
stimulus.Habitusemphasizesdexterity,knowhow,practicalreason(Bourdieu1984;Collet
2009;Crossley2013;Lindbladh&Lyttkens2002).

ForBourdieuhabitus(1984)hastwofaces:onesideshowsitselfasagenerativeprinciple
ofjudgmentandontheotheritisthesystemofclassification.Thehabitusisastructuring
structure, which organizes practices and the perception. Habitus is the ability to generate
classifiable practices and works, and the faculty to differentiate and appreciate these
practices and products (taste) which form the represented social world. In other words,
habitus is a disposition within the individual to make specific decisions and to have
particular perceptions, retaining individual agency while following a system of rules and
constraints (Bourdieu 1984; Collet 2009). The sociological perspective, and particularly
Bourdieus,shareoneaspectwiththeDispositionalPerspective(DIS).Thisistheideathat
habitsinformbeliefs,deliberationandactionsiteratively.

The rigid distinction between empiricism and transcendentalism, habit and habitus,
accordingtoFoucault(1973)cannotbesustainedanddoesnotneedbesustainedasthey
tendtoconverge(Crossley2013).Inthebeginningofthesetwoparadigmsonewouldsee
humansasbiologicalmachines,theproductofpureevolutionwherereasonwouldrespond
homogenouslyalongtime.Theotherwouldgiveabsoluteagencytotheperson;opposethe
attemptstopredictbehaviorandtreatindividualreasonasheterogeneousalongtime.
30

This thesis rejects the mechanical automaticity of the behaviorism and acknowledges the
human capacity to act freely. However, it also recognizes the value and plausibleness of
modeling behavior and predicting it. This thesis then stands for an emerging convergent
perspectivedescribedbyCrossley(2013):Developmentsinhumansciencehaveincreasingly
converged with arguments regarding mind and action in philosophy. The mechanical
worldviewdoesnotremaininmuchofnaturalsciencesorintheempiricalsocialsciences.It
isnowpossibletodevelopaphilosophicallysophisticatedandacceptablemodelofhuman
actionandexperiencewithinanaturalisticframework.

Inthisperspectivetheempiricisthabitandtranscendentalhabitusaretakenintoaccount
to understand the holistic nature of habits. For instance, in the next section, habits are
acknowledgedasautomaticlike,notasautomaticormechanicallyautomatic.Thisaccepts
thathabitstendtorepeatheuristicallyinthepresenceofcontextualcues,buttheycanbe
discontinuedandredirectedbypayingattentiontoaction.

2.6 AttributesofHabit

2.6.1 Habitsareacquisitions(learned)

Habits are different to instincts and other reflexlike or automaticlike actions in that it is
learnt (Hodgson 2010). Besides, habits represent the knowledge accrued across multiple
past occurrences (Wood & Neal 2009). Theory on habit posits that a condition required
developingahabit.Theconditionistherepeatedperformanceofbehavior(Ronis,Yates&
Kirscht 1989). However, satisfaction with consequences of the performed behavior and a
stablecontextarealsonecessary(Aarts&Dijksterhuis2000;Bargh1990;Limayem,Hirt&
Cheung 2007; Ouellette & Wood 1998; Sheeran et al. 2005; Verplanken & Aarts 1999;
Wood,Tam&Witt2005).Werefertocontextastheenvironmentwherebehaviortakes
place.Thismayincludethephysicalenvironmentandinfrastructure,butalsospatial,social
andtimecueswhichinstigateaction(Verplankenetal.2008,p.122).

Learningahabitisagradualprocessofassociationbetweenresponsesandthefeaturesof
thecontext,WoodandNeal(2009)highlighttheimportanceoftherewardingresponsein
theprocessofinstillingahabit.
31

The original behavioral performance will tend to have a greater level of intentionality. As
the habit is learnt, a lower level of awareness is needed to perform habitual behaviors
(Danner,Aarts&deVries2008).Hodgson(2010)acknowledgesBeckers(1992)positionas
validinthatrationalchoicescanleadtotheformationofhabits.However,Hodgsonpoints
outthatrationalchoicesthemselvesarealwaysreliantonpriorhabits.

2.6.2 Habitsaretendencies(predictable)

Habits create predictability (Loibl, Kraybill & DeMay 2011). Although there are different
perspectives on why, how and to what extent habits can predict future behavior, most
literaturepresentsacausalrelationbetweenpasthabitualbehaviorandfuturebehavioror
habits and future choice, habit trained capacities and rulelike propensities of behavior
(Bourdieu1984;Hodgson2010;Ouellette&Wood1998;Triandis1977).

Thecompoundofallhabitsofapersonaffectindividualbehaviorinthesensethatrational
choices themselves areat all times and inevitablydependent on former habits. When
peoplearedistracted,theytendtodecideandactbasedonhabits,andtheymaynoteven
recognize information relevant to an alternative behavior. In that way habit also affects
choiceinacontinuouscycle(Becker1992;Danner,Aarts&deVries2008;Kremers,vander
Horst&Brug2007;Limayem,Hirt&Cheung2007;Marchal2010;Ouellette&Wood1998;
Ronis,Yates&Kirscht1989;Verplanken&Aarts1999;Wood&Neal2009).

Differences on perspective arise only on why, how and to what extent habits can predict
future behavior. On one hand there is a behavioristic perspective which is inclined to
consider more mechanical correlations between past and future behaviors. In that way,
habit would predict future behavior because once a stimulus associated with a response
appearstheresponsefollowsaccordingly(Becker1992;Hull1943;Watson1998).

On the other hand, an individual is considered as an agent who responds and adapts to
solicitationsandconstraintsoftheexistingenvironment.Buthavingtrainedcapacitiesand
imbibed structured propensities, individuals appear to act as following rules while they

32

retaintheiragency(Bourdieu1984;Bourdieu1985;Bourdieu2008;Bourdieu&Wacquant
1992;Collet2009;Wacquant2005).

2.6.3 Habitsarepatterns

Individuals repeatedly tend to perform the similar behavior under similar circumstances.
This characteristic suggests that when habits have been formed, subsequent behavior is
awakenedbyspecificenvironmentalcues(Kremers,vanderHorst&Brug2007;Limayem,
Hirt&Cheung2007).Personsformhabitsbyrecurrentlypursuinggoalsthroughparticular
means in stable contexts. They tend to repeat those actions which are gratifying in some
way or those which produce them valued outcomes (Wood & Neal 2009). Repetition is
thenperhapsthe first element which causes habit to be patterned. Then, association
betweenenvironmentandsubsequentbehaviorwouldprobablybethesecond.

ForBourdieu(1984),repetitionwouldnotbeindividualsjustreproducingpastbehavioror
following behavioral rules. However, it is not that Bourdieu denies that people do things
whichareverysimilarifnotidenticaltowhattheydidbefore.Collet(2009)presentshow
Bourdieusemphasisisplacedinthehumancapabilitytotakeintoaccounttheevolutionof
thesocialfieldsinwhichtheyare,andusetheiragencyinsimilarsituationswhicharenewin
time. Habits are not just patterns of behavior, but the structure that creates patterns.
Bourdieu (1984) expresses this idea in a phrase: The habitus is not only a structuring
structure,whichorganizespracticesandtheperceptionofpractices,butalsoastructured
structure(p.170).

In the discipline of economy the habitual pattern is expressed as stochastic and

deterministic elements (Hodgson 1997). Covariance between the cues in the context and
behaviorrepresentthedeterministicsideofapattern.Still,thereisnocompletecorrelation
between inputs and outputs, stimuli and response in the habitual patternat least not
always.Insomecasesthepreferredselectionsofthechooserwillbeinconsistentwitheach
other(Becker,DeGroot&Marschak1963).Socialsciencehasnameditagency(Bourdieu
1984), as opposed to mainstream economics where this unpredictable component of the
patternistreatedasastochasticelement(Hodgson1997).

33

2.6.4 Habitsareextrapolators

Bourdieudescribeshabit(habitus)assystemsofdurable,transposabledispositions(1992,
p.53).Ahabitthathasbeenacquiredinaparticularcircumstancecanbeappliedbeyond
thelimitsofwhathasbeendirectlylearnt(Bourdieu1984,p.170).Habitsareshapedbythe
possibilities and impossibilities, freedoms and necessities, opportunities and prohibitions
inscribed in the objective conditions. Accordingly habits are objectively compatible with
theseconditionsandinasensepreadaptedtotheirdemands(Bourdieu&Wacquant1992).
Habits, as systems of generative schemes, can simply be transferred to the most varied
areas of practice when applicable (Bourdieu 1984). Thus, habit is useful to deal with
uncertainty,complexityandchange(Hodgson2010).

2.6.5 Habitstendtoberewarding

Habitsaredevelopedasbehaviorsthatseemtobesatisfactoryinachievingsomegoalsuch
asdrivingacartoadestinationoreatingfoodforthesakeofpleasure(Verplanken&Orbell
2003). Satisfactory experiences after behavior are fundamental for habit development as
they raise the tendency to repeat a given course of action (Aarts, Paulussen & Schaalma
1997;Limayem,Hirt&Cheung2007).

Aarts and Dijksterhuis (2000) consider Habits to be the link between goals and behaviors,
and they may be functional in obtaining certain intended goals or end states (Aarts &
Dijksterhuis2000;Aarts,Verplanken&Knippenberg1998;Limayem,Hirt&Cheung2003).
GoalDirected behavior [] does not [explicitly] mean consciously decided or planned
(Guinea & Markus 2009, p. 434). Therefore less conscious goals might intervene in
behavioralreplication,evenwhenbehaviorfailstoachieveconsciouslysetgoals.Hodgson
(1997, p. 665) explains the difference between the effectiveness in pursuing a conscious
goalandthefulfillmentofwellbeingbehindhabitualrepetitioninthefollowingexample:

Clearlysomehabitsorrulesareefficaciousandothersarenot.Somerulessuch
aswhentragedystrikes,sacrificeafavoredanimaltoplacatethegodsmayhave
no scientific foundation. However, the association of ritual sacrifice with

34

 subsequentwellbeingisconsistentwithasystemofbelief,andrecoursetothe
ruleisthusexplicable.

This suggests that habits are repeated because they are effective to provide a reward
(pleasure, wellbeing, comfort, etc.), which can be found or not and even opposed to
achievingconsciouslysetgoals.Inparticular,whenweconsiderhabitsthatareunwanted
[] it is important to realize that, from the individual's perspective, such a habit is
functional, and thus "wanted", in achieving some goal, (such as feeling comfortable)
(Verplanken&Aarts1999,p.106).

2.6.6 Habitsarelatentuntilactivated

Onceahabitislearntitremainsdormantwithinthepersontobeawakenedincorrelation
with specific cues of the context. However, there is controversy about volition and
intentionalitywhileactingbyhabit.Astreamhasconsideredhabitstobenonvolitionaland
unintentional, but the other stream posits that automatic or routinized actions can be
volitionalandpartofanintentionalbehaviorsystem(Ajzen2002;Ouellette&Wood1998;
Polites2005).

Inregardstotheoriginofchoicetwostreamswereidentified.Ononehandastreamwas
identifiedwhichconsidershabitasasequentialcorrelationinbehavior(Hodgson2010).This
streamisreferredhereasUniformBehaviorParadigm(UBP).InUBPhabitisdefinedasa
positive relation between past and current consumption where past tends to be uniform
withthefuturebehavior.

Ontheotherhand,asecondstreamwasidentified.Formerepracticalreasonsitwasnamed
Dispositional Paradigm (DIS) in this review. DIS sees habit as dispositions or submerged
repertoires of potential behavior that in the future might be executed or not. In contrast
withuniformbehaviors,DISembraceschoiceinsteadofseeingitasanindependentcausal
power.

AccordingtoHodgson(2010),whocanbeidentifiedwithDIS,thereisacausalitydilemma
implied in UBPs conception. He explains that such paradigm fails to explain the origin of
35

individual choice. This is because habit determines choices, and choices determine habits.
Hodgson(2010)(DIS)settlesthedilemmabysuggestingthathabitderivesfromchoiceand
instinct,andnotonlyfromchoice.

EconomistsconsideredmainstreambyHodgson(2010)assuchastheBritishLionelRobbins,
and the Americans Gary Becker and Kevin Murphy would match the profile of the first
paradigm presented as a uniform behavior, empiricist paradigm. Psychologists,
philosophers, and economists such as the Americans William James, John Dewey, Torsten
BundeVeblen(NorwegianAmerican)andtheBritishGeoffreyM.Hodgson,canbeidentified
with schools of thought such as Pragmatism and Evolutionary Economics (heterodox
economics),matchingtheprofilepresentedforaDispositionalParadigm(DIS)(seeTable2.4
MapofPerspectivesofHabit).

2.6.7 Habitsareautomaticlike

In Psychology and Information Systems, automaticity has been regarded as the main
characteristicofhabit,howeverthatisnotnecessarilyconsistentwithabroaderspectrum
of disciplinesat least not at the level of the definition of the term. For instance in a
literature review for a study on Information Systems, Limayem et al. (2007) report 43
explicitandimplicitdefinitions ofhabit,48% ofthemfocusonautomaticity,whileinthis
thesis definitions sample (see Section 2.4 The concept of habit) only 9.6% give central
attentiontoautomaticityasanessentialcharacteristicofhabit.Itwassurprisingthatnone
ofthedefinitions,inthedefinitionssample,classifiedashumanities(artsandhumanities,
ethics, literature, music, philosophy, religions and theology, and semiotics) include
automaticityintheircoredefinitions.

This study does not disregard the importance of the automatic effect of habits, however
nowadayssomeauthorsconsiderhabitualbehaviorisnotanautomatictendency(Pahnila,
Siponen&Zheng2011,p.23);becauseafterallmosthabitsreflectchoicesthathavebeen
foundsatisfactoryinthepast(Verplanken,Aarts&VanKnippenberg1997,p.558),which
increases the inclination to repeat such behavior in the future (Pahnila, Siponen & Zheng

36

2011). Bourdieu also emphasizes that the process of habitus occurs far from mechanical
determination(Bourdieu1984).

Barghs (1989) approach to automaticity acknowledges the intervention of different levels

ofattention,intentionandawarenessinthreedifferentformsofautomaticity.Herefersto
theautomaticeffectswhichfallinthosethreeclasses.Therefore,inacknowledgmenttothe
diverse possibilities of consciousness, control or agency held upon individual behavior;
automaticityisherepresentedasanautomaticeffect.

2.6.8 Habitsareefficient

Behaviors driven by habit can be performed easily in parallel with other actions with
minimal attention (Chen & Chao 2010; Gu et al. 2010; Ouellette & Wood 1998). Mental
efficiency implies saving memory space and processing time while performing complex
sequences of actions in a frequent way (Polites 2009). This also involves less thought as
controlofthebehavior(Danner,Aarts&deVries2008;Lallyetal.2010;Marchal2010).
Empirical research also provided evidence that habitual behaviors are less complex than
nonhabitualbehaviors(Lankton,Wilson&Mao2010;Wood,Quinn&Kashy2002),

Thereisdisagreementtosomeextentontheamountofinformationtakenintoaccountto
makehabitualchoices.Ononehanditisassumedthateverypieceofinformationistaken
into account while acting guided by habits. The first assumption is consistent with
paradigmsofuniformbehaviorandbehaviorism.Whereasanopposingperspectiveaffirms
that such assumption implies a storage capacity problem considering the notions of
boundedrationalityintroducedbySimon(1997).Hodgson(2010)alsoaffirmsthatahabitual
system encapsulates past adaptive behavior in a way that does not require all pieces of
informationberetained.Thissecondpostureisalignedwithauthorspreviouslyidentifiedin
theDispositionalParadigm(DIS)(seeTable2.4MapofPerspectivesofHabit).

2.6.9 Habitsareshared(social)

Habitsarenaturallysituatedinthestructureofsociallife(Rozin2001;Wood,Quinn&Kashy
2002),thereforesocialperceptionsinfluencehabitsandhabitsinfluencesocialperceptions.
37

In social structures habits are shared schemes of thinking and behaving (Alakrpp et al.
2010; Bourdieu 1984; Louis & Sutton 1991). While belonging to a group with a common
context habit (habitus) will tend to generate what it is considered reasonable and
common sense within the limits of such a context or objective regularities (Bourdieu &
Wacquant 1992, p. 5556). People have routinized social lives and respond to social cues
(Orbelletal.2001).

More importantly, habits are somehow transferable in a social structure. When an

individual has little or no experience with a proposed behavior, the social influence may
helptheindividualinhisorherdecisionmakingprocess(Pahnila&Warsta2010,p.625).
Later on recommendations from others can strengthen a habit too (Aarts, Paulussen &
Schaalma1997).

2.6.10 Habitsareunique(individual)

Despite being inserted extensively within the collective system, habits are generally
acknowledgedasastructurethatexistsattheindividuallevel(Becker2005;Hodgson1997;
Polites2009;TorresMaldonadoetal.2011).Aparticularcompoundofhabitsistheresultof
individual history of behavior which is unique for every individual. Therefore, habits are
uniqueandidiosyncratic(Danneretal.2011),andhabitsreflectapersonssenseofidentity
(Norman 2011; Trafimow & Wyer 1993). Lankton et al. (2010) suggest that habitual
behaviorsaremoreselfidentifyingthannonhabitualbehaviors.

The sense of identity derived from the factual uniqueness of a persons habits has raised
some disagreement in the literature. The SelfReported Habit Index (SRHI) (Verplanken &
Orbell 2003) included the identity or personal style as it was considered as an important
element of theory of habit. However, while discussing measurement of habit as a
psychologicalconstruct,otherauthorsstatethatselfidentityisnotanecessarycomponent
ofhabit(Gardneretal.2011;Sniehotta&Presseau2012).

Sniehotta & Presseau (Sniehotta & Presseau 2012) argue one would be hardpressed to
suggestthathabitualeatingofpotatochipsispartofonesselfidentity(p.139).However,

38

thechipsandtheiravailabilitydoescontaintheparticularstyleofthecultureandsettingin
whichtheywereproducedandprocessed(Bijker,Hughes&Pinch1987).Thereforechips
maycontainseveralstylisticelementswhichtheindividualmayidentifywithhimself.

2.6.11 Habitsareelastic(resilient)

The more situation and action are associated; the stronger habit becomes. While the
association between a context and a performance grows, a recurrence of the behavioral
activation increases progressively in the given circumstance. This is consistent with the
cognitivemotivational standing point (Aarts & Dijksterhuis 2000; Sheeran et al. 2005;
Verplanken2006;Wood&Neal2007)asdescribedbyPolites(2009).

Thestrengthofahabithasbeenregardedasafunctionofrepetitionforaslongasthereisa
rewardassociatedwiththeactionwhichfollowsthecontextualcue(Hull1943).Inregards
totherationalistperspectiveofbehavior,Ajzen(2002)andTriandis(1979)wouldapparently
not conflict with this proposition. Therefore, the strength of habit can generally be
consideredasdeterminedbythefrequencyofperformanceinthepastwhileitoccurredina
similar context (Danner, Aarts & de Vries 2008; Lally et al. 2010; Ouellette & Wood 1998;
Pahnila&Warsta2010).

2.6.12 Habitsareplastic(malleable)

Paradoxically, habits are elastic as well as plastic. Habits are resistant to change, but they
are also susceptible to change. Wood and Neal (2009) explain that much important
knowledge would be at risk if the information accrued over a long period of time could
easilybeeradicated.Thereforechangeoccursslowlyoverrepeatedexperiences.

Changeinhabithappensinparticularconditions.Habitcanbebrokenwhenchangesinthe
contexttakeplace(Verplanken&Wood2006).Besides,ithasbeenobservedthatoneofthe
mostsuccessfulstrategiesforstoppingundesiredhabitsisvigilantmonitoringandattention
ontheunwantedresponse(Wood&Neal2009).Incomplexcircumstances,wherenatural
and social environment is inconstant the modificatory power of habits, plasticity becomes
evenmorevitalthaninstincts(Hodgson2010).
39

2.6.13 Habitsareknowable(susceptibleofmetacognition)

Habitshavebeensaidtobeunconscious.However,theuseofthistermcanbeconfusingas
itcanmeanthatwhenactingoutofhabit:1.thepersondoesnotknowwhatheisdoing
(see Mittal 1988, p. 998); 2. The person is not rationally verbalizing a logical discourse
(verbalizingcanbeatthemindlevel,seeLindbladh&Lyttkens2002);3.Thepersonisnot
aware of his past behavior or behavioral pattern; 4. The person does not know that his
presentactionsarecreatingahabit.

Literature suggests that the individual is unaware of the situational trigger leading him to
performanaction,unawareofhowthetriggerisinterpretedatthemomentitoccurs,and
unaware of the factors responsible for his habits. However, the individual is aware of the
effect of the stimuli (Bargh 2002; Bargh et al. 2001; Fine 2008; Guinea & Markus 2009;
Polites2009;Wood,Quinn&Kashy2002).

Mittal(1988)explainsthatforanactiontobeassumedunaware,aplanorverbalizedself
instructionsmustbeabsentatthemomentofaction.Thatwouldimplyifhabitualactions
wereunconsciousinthatwaythattheindividualisnotawareofahabitdrivenactwhileit
isunfolding.Mittalsustainsthat,fortheindividual,beingawareofhisactwouldequalto
beingawareofhisroleaswillingexecutorandthereforeofhisintentiontoexecutetheact.
Whileovertlypresentinganassumptionofthisfirsttypeofunawarenessinhisstudy,Mittal
pointsoutthatanindividualcanretrospectivelybecomeawareofhisactsbythetracesof
hisunawareactions,particularlyinthecaseofthoseconsideredhabitdrivenacts.

2.7 Prototypedefinitionofhabits

Theobjectiveofprovidingaprototypedefinitionforhabitsistoexplainaproblemthatwas
identifiedascommoninmostofthedefinitionsofhabit.Thisprototypedefinitioniscalled
suchbecauseitmightbefarfromclaimingtobeanultimatedefinition,andbecauseitisnot
directlyusedtoinformthemeasuresofhabittechnologyfit.Instead,thisdefinitionaimsto
illustratewhyresearchmightstillbecallingforbetterdefinitionsandoperationalization(see
Crossley 2013; Lally et al. 2010; Limayem, Hirt & Cheung 2007; Saba & Di Natale 1998b;

40

Triandis1977;Tuorila&Pangborn1988).Italsoaimstoofferaholisticperspectiveofthe
concept, which is frequently incomplete in operational definitions. Even though this
prototype definition does not directly inform the measures of habittechnology fit, it
significantlydefinestheapproachtomeasuringhabitsinthisthesis.

Theconceptofhabitbelongstotheeverydaylanguage,whereitsdefinitionisvariableand
imprecise(Crossley2013).Thisimpliesadirectproblemtooperationalizehabits,becausea
researcherunderthepositivistparadigmmayneedhighlevelsofspecificityinthedefinition
in order to apprehend the reality (Blaikie 2010; DeVellis 2012; Guba & Lincoln 1994).
However,thisproblemmightfindasolutioninfutureresearchifitisempiricallyapproached
byontologicalanalysistechniques.

Ontology,isadisciplineofPhilosophythatdealswithwhatis,withthekindsandstructures
ofobjects,properties,andotheraspectsofreality(Welty&Guarino2001,p.51).Research
has advanced in developing methods and tools applied to knowledge systems with
foundations on Ontology. According to Welty and Guarino (2001) in order to define a
concept, its identity, essence, unity, and dependence should be established. Identity deals
with differentiation between instances from other instances, it refers to characteristic
propertieswhichmaketheinstanceuniqueasawhole.Unitydenotesthewaysinwhichthe
parts of the instance are bind together without anything else. Essence is informed by
identity and unity, but it is also related to how an instance can be reidentified over the
passageoftime.Finallytheontologicaldependencereferstointrinsicorextrinsicrelations.
Intrinsic properties are not dependent on other entities, whereas the extrinsic properties
areassignedbyexternalagents.

Theseontologicalconceptsputforwardquestionsthatrequireclearanswersbeforehabits
can be properly defined. For example, every time a question is asked, such as, is that a
habit? The researcher should be able to systematically establish the identity of the
observed,andprovideabinaryanswer(yesorno).IfitwasaskedIsthatacomponentof
habits? In the same way, the researcher should have the elements to judge. After time
passed,somethingthatwasidentifiedasahabitcanitstillbeidentifiedassuch?

41

Thetermhabitbelongstotheeverydaylanguage(Crossley2013).Therefore,itmayrequire
consideringcontextsto besystematicallydefined,asisdoneinsemanticsystems(Moore,
Evans&Tadros2013).Identifyingwhichofthepropertiesdescribedforhabitareintrinsicin
common life contexts and discriminating them from those which are assigned by agents
(suchasresearchers)mightbenecessarytodeliveranadequateconceptualization(Gruber
1993;Welty&Guarino2001).

Section 2.4 The concept of habit, analyzed a sample of 135 definitions from multiple
disciplines.Themostfrequentoneworddefinitionsforhabit(orderedbyfrequency)were:
tendency,pattern,behavior,ways,acquisition,response,androutines.Thesesevenwords
arenotfromthesamesource.Indeed,eachwordisthecoretermfromonedefinitionofthe
135. These words are used in academic papers of several disciplines and dictionaries (see
Appendix4andAppendix3).Therefore,theymayreflectitisusedinthecontextofsocial
andcommonlife.However,KlcknerandMatthies(2004)disapprovetheuseofhabitand
routine as synonyms. Limayem, Hirt and Cheung (2007) discriminate habit from behavior
androutine,astheyareconsideredproxieswhichdonotcompletelyrepresenthabits.The
contradiction suggests all these terms have a relationship with habits, but hardly any of
themcanfullyrepresentthecomplexconceptofhabits.

Section 2.6 analyzed peerreviewed literature of habit, and then presented a synthesis of
the attributes of habits. This time the analysis originated from profuse explanations of
habits, rather than succinct definitions. On those bases, this thesis posits a prototype
definitionforhabit.Thisdefinitionisnottheresultofontologicalanalysis.Howeveritwas
guidedbythebasicaspectsofontology,suchasidentity,essence,unity,anddependence.

Habits are a complex system of structuring structures which fulfill the following
characteristics:Theymustbeacquiredbylearning,predictable,patterned,rewarding,elastic
butplastic,uniqueinindividualsbutsociallynested,cognitivelyefficientandautomaticlike,
their execution must require minimal awareness and control, but still be retrospectively
knowable by the traces of action. Habits must also be susceptible of extrapolation to new
situations,andlatentuntilactivatedintheformofbehaviorsorthoughts.

42

This prototype definition is comprehensive but impractical. It contains thirteen elements,

which could be broken into several dimensions each. This prototype definition may offer
improvement to the ontological unity, compared to other definitions of habit in the
literature. However, it might be too complex to be operationalized directly. This inherent
complexityoftheconceptmayexplainwhycallsforbetterdefinitionsofhabitcontinuedto
rise(seeCrossley2013;Lallyetal.2010;Saba&DiNatale1998b;Triandis1977;Tuorila&
Pangborn1988).

Inordertoaddressthedifficultiesinherenttothehighlevelofdimensionalityofhabit,this
researchwillreviewpreviousapproachestooperationalizationofhabits.Itwillexplorethe
possibilities offered by the personenvironment fit literature (Section 2.12 ). From the
theory on personenvironment fit, it might be worthwhile to anticipate the potential
adequacyofperceivedfittoaddressthesedifficulties(seeSection2.15).Whenmeasured,
perceived fit allows respondents a complete cognitive manipulation of their evaluation.
Individuals are allowed to define the salience of the various dimensions of the variable in
question (KristofBrown & Billsberry 2012; KristofBrown, Zimmerman & Johnson 2005;
Kristof1996).Thisthesisusesthepropertiesofperceivedfittocapturethehabitswhich
are most important to the individual. However, it may also serve to capture the salient
dimensionsofhabititself.

2.8 ClassificationsofHabitinLiterature

Althoughoccasionallydifferenttypesofhabitsarementionedinliterature,classificationsof
habitsfoundwerequitelimited.Thus,acompilationofclassesofhabitsispresentedinthis
review. When habits were mentioned in literature as a type, they were collected and
organized. The output of the grouping procedure showed that habits can be classified at
leastbytheirlevelofintentionality,morality,observability,propagationandplasticity.

43

2.8.1 Bylevelofintentionality(intentionalandunintentional)

Dewey (2002, p. 28) has provided elements to define a class based on its level of
intentionality. He considers the existence of habits of two opposite kinds, intelligent and
routine.Thedifferencebetweenthemreliesontheextentofthemechanicaleffectofthe
actionversusthedegreeofartisticskill.

Bargh (1989, p. 7) refers to the same classes as preconscious, postconscious and goal
dependent automaticity. The first refers to those actions occurring before conscious
awareness,thesecondtothosewhichrequiresomeconsciousprocessingbutstillproduces
an unintended outcome. In contrast goal dependent automaticity would require full
intentionality.

2.8.2 Bymoralquality(goodandbad)

Habitscanbegroupedbytheirlevelofmorality(goodandbad).Dewey(2002),presentsthe
notionofgoodhabitsandbadhabitswheretheformerarevirtuesandthelatterarevices.
Extensive literature makes reference to this classes of habit, for example Hodgson (1997)
Loibletal.(2011)Klckneretal.(2003)Wood andNeal(Wood&Neal2009)SabaandDi
Natale(1998b)VerplankenandFaes(1999)tonamesome.

2.8.3 Bylevelofvisibility(observableandhidden)

Theextentofobservabilityservestodescribeafewclassesofhabitswhichinsynthesisare
explicit habits and implicit habits. Whereas habits of locomotion (motor) and verbal are
explicit because they are clearly observable; habits of mind or thought, feelings, belief,
judgment, desire, emotion, and perception belong to the category of implicit habits
hiddenfromdirectobservation,stillobservableusingtechnicalprocedures(Bourdieu1984;
Dewey2002;Staats1959,Louis,1991#3033;Veblen1898;Watson1998).

44

2.8.4 Bythelevelofcommonality(individualandcollective)

Habits can be classified as collective habits or individual habits. In a social system,

individuals share cognitive structures that guide their interpretations and behaviors and
amonggroupstherearedifferentcollectivehabitsofmind(Alakrppetal.2010;Louis&
Sutton1991).Deweyalsomakesreferencetocollectivehabitswhenhestates:Forpractical
purposes morals mean customs, folkways, established collective habits (2002, p. 30).
Bourdieu(1992)heldthathabitusisasocializedsubjectivity(p.126).Ontheotherhand,
habitisacknowledgedasastructurethatexistattheindividualleveldespitebeinginserted
extensivelywithinthecollectivesystem(Becker2005;Hodgson1997;Polites2009;Torres
Maldonadoetal.2011).

2.8.5 Bythelevelofplasticity(rigidandflexible):

Habits can also be categorized as rigid and flexible. Dewey (2002) refers to rigid habits
which insist upon duplication, repetition, [and] recurrence (p. 89). He remarks that the
levelofplasticityisvariablewhichmakessomehabitsrigidandothersthelessrigidmightbe
namedflexible.

2.9 Adistinctionbetweenhabitvsinstinct

Whentheoryonhabitisexplainedtheconceptofinstinctsusuallyarises.Habitandinstinct
are highly similar in their characteristics and functions. They differ mainly in the acquired
natureofhabitandbiologicallyinnatecharacterofinstinct.Instinctisthefirstlinkinachain
ofdevelopmentwhichderivesintohabitandjudgment(Hodgson2010).However,instincts
arenotgenerallyconsideredasaclassofhabit,asMargolis(1990p.29)suggests:

Habits must be built out of instincts; judgment must somehow derive from
instinctsandhabits.Sooneoftheusefulwaystocategorizebrainswouldbeto
distinguish among those that work on instinct alone, those that use instinct +
habit,andthosethatuseinstinct+habit+judgment.

45

This section presented ten classes of habit (intentional and unintentional, good and bad,
observableandhidden,individualandcollective,rigidandflexible)groupedbytheirlevels
of intentionality, moral quality, visibility, commonality and plasticity respectively. Scarce
work has been done in previous research in order to systematically classify the existing
types of habits. Most authors just occasionally mention some of types. The present study
will not deepen in this, but it is acknowledged that future research should extend the
categorization of habits, including levels of awareness, efficiency and control. This thesis
provides initial identification of classes of habit, which constitutes a contribution to
literatureofhabit.

Thefollowingsectionpresentsempiricalevidenceoftherelationbetweenhabits,behavioral
intentionandactualbehavior.Measurementsofhabitarediscussed,andagapinthehabits
thataremeasuredisidentified.

2.10 HabitandBehavior

Twotypesourcesindicatetherelationofhabituponbehavior,thosewhichpresentapurely
theoreticalperspectiveand thosewhich include empirical evidence. There has been some
discussionabouttheeffectsofhabitinrelationtobehaviorandintention.Mostliterature
inclinestosuggestthathabitmightnotinfluenceintention,giventheautomaticlikeeffect
ofhabit.However,empiricalevidencesuggestssomethingdifferent.

This review found 54 articles empirically testing the relationship of habit upon behavioral
intention and behavior (see Table 2.5). A direct relationship between habit and actual
behavior (HAB) was observed 36 times. Closely in second place, 28 articles report on
(HBI).Only11articlesreportonhabitasamoderatorofintentionandbehavior.

46

 EMPIRICALEVIDENCEOFHABITUPONBIANDAB
Total Not Significant Notsignificant
Relation Reported relation
Reports reported (P<.05orp<.001)
relation
HBI 28 51.9% 26 48.1% 27 96.4% 1 3.6%
54 HAB 36 66.7% 18 33.3% 31 86.1% 5 13.9%
HxBIAB 11 20.4% 43 79.6% 8 72.7% 3 27.3%
1 (EscobarRodrguez & CarvajalTrujillo 19(deBruijnetal.2009) 37(Verplanken&Orbell2003)
2013) 20(Gardner2009) 38(Orbelletal.2001)
2(Han&Farn2013) 21(Limayem&Cheung2008) 39(Saba,Vassallo&Turrini2000)
3(Huang,Wu&Chou2013) 22(Wu&Kuo2008) 40(Trafimow2000)
4(Kangetal.2013) 23(DeBruijnetal.2007) 41(Verplanken&Faes1999)
5(Klckner2013) 24(Limayem,Hirt&Cheung2007) 42(Ouellette&Wood1998)
6(Nikou&Bouwman2013) 25(Liao,Palvia&Lin2006) 43(Saba&DiNatale1998a)
7(Raman&Don2013) 26(Thgersen2006) 44(Saba&DiNatale1998b)
8(Tseng,Chang&Woo2013) 27(vanEmpelen&Kok2006) 45(Sabaetal.1998)
9(Venkatesh,Thong&Xu2012) 28(Verplanken2006) 46(Verplanken,Basetal.1998)
10(Barnes2011) 29(Honkanen,Olsen&Verplanken2005) 47(Bergeronetal.1995)
11(Chen&Lai2011) 30(Kim&Malhotra2005) 48(Towler&Shepherd1992)
12(Loibl,Kraybill&DeMay2011) 31(Wood,Tam&Witt2005) 49(Ajzen1991)
13(Norman2011) 32(Gefen2003) 50(Montano&Taplin1991)
14(Pahnila,Siponen&Zheng2011) 33(Klckner,Matthies&Hunecke2003) 51(Bagozzi&Warshaw1990)
15(DeBruijn&Rhodes2010) 34(Limayem&Hirt2003) 52(Charng,Piliavin&Callero1988)
16(Guetal.2010) 35(Limayem,Cheung&Chan2003) 53(Mittal1988)
17(Lankton,Wilson&Mao2010) 36(Limayem,Hirt&Cheung2003) 54(Wittenbraker,Gibbs&Kahle1983)
18(DeBruijn&VanDenPutte2009)
Table2.5EmpiricalEvidenceofHabitUponBIandAB (Source:Author)
(SeeAppendix5)

Empirical evidence has shown instances where habit has an impact upon behavioral
intention (HBI), actual behavior (HAB) or moderates the relation between them
(HxBIAB).Althoughthenumberofstudiesisrelativelysmalltogeneralize,itshowsthat
habit as a moderator was not found significant 3 out of 11 times. Habit as a direct
determinant of actual behavior was found not significant in 5 out of 36 observances. The
studieshypothesizinghabitasadeterminantofbehavioralintention(HBI)onlyfailedto
besupportedinonly1in28occurrences.

The only occurrence where the hypothesis (HBI) was found not significant, was a study
conductedbyRamanandDon(2013).ThestudywasbasedonUTAUT2(Venkatesh,Thong&
Xu2012)andanalyzedwithPLS.ItstudiedtheuseofaLearningManagementSystemusing
asampleof320usefulresponsesprovidedbystudentsinMalaysia.

In contrast to this one study, other two articles on UTAUT2 reported on habit and
behavioralintention(HBI)Findingitsignificant.AllthreestudiesusedLimayem,M.,Hirt
and Cheungs (2003) measurement items, and all tested (HBI) and (HAB). A study on

47

MobileInternet(Venkatesh,Thong&Xu2012)usedPLStoanalyzetheusefulresponsesofa
1360adultssample.Anotherstudyononlineairlineticketpurchasing(EscobarRodrguez&
CarvajalTrujillo 2013) used SEM and a sample of 1512 users. In the two cases both
hypotheses were supported. In contrast, the study on Learning Management Systems
(Raman&Don2013)usedasampleof320studentsandbothhypotheseswerefoundnot
significant.

In 5 cases the hypothesis (HAB) was found not significant. Nikou and Bouwman (2013)
studiedMobileSocialNetworksusingasampleof336respondentsinChina.Datafromthe
study was analyzed with SEM. Raman and Don (2013) used the theoretical framework of
UTAUT2tostudyLearningManagementSystems.Datawasobtainedfrom320Malaysian
students and analyzed with PLS. Lankton, Wilson and Mao (2010) studied University
Internet Applications including 371 undergraduate students in the survey. Their data was
analyzed with PLS and ANOVA. Montano and Taplin (1991) conducted a study on
Participation in Mammography using the Theory of Reasoned Action. Data was obtained
from 946 women above 40 years old, and it was analyzed using multiple correlations.
Bagozzi and Warshaw (1990) studied Losing Weight in a sample of 240 undergraduate
students.Datawasanalyzedusingmultipleregression.

In three cases the hypothesis (HxBIAB) was found not significant. Han and Farn (2013)
conducted a study on Pervasive Business Intelligence Systems reaching 117 students
through snowball sampling. The analysis method used was PLS. Gardner (Gardner 2009)
researchedTravelModesbasedontheTheoryofPlannedBehavior.Thesampleconsisted
of107staffandstudentcarcommuters,anddatawasanalyzedwithregression.Wood,Tam
andWit(2005)alsoanalyzeddatawithregressionon115students,usingsamplesomehow
similartothefirsttwostudies.

The reasons for failure are a limitation for pure quantitative studies. In this review, the
commonalitiesrelatedwiththeoreticalframework,measurement,sampleoranalysistype,
werenotenoughtosuggestapattern.However,habithasasmallerfailureratepredicting
intention;thisresearchconsidersitasafirstreasontoproposesuchhypothesis.

48

Theempiricalevidencepresentedinthepreviousparagraphs,suggestthatotherhabitsmay
also have a significant influence on behavioral intention. This thesis uses this evidence to
arguethathabittechnologyfitmayhaveasimilarrelationshipwithbehavioralintention.

2.11 Measurementofhabits

Until now, the main approaches to operationalize habit are: Frequency of Behavior, the
Response Frequency Measure, and the SelfReport Habit Index (SRHI). However, these
approaches focus on specific habits which are usually related to the target technology
directly.

Frequency of Behavior (FB) refers to a measure of objective or selfreported regularity in

individualperformanceofonetypeofaction(Ouellette&Wood1998;Triandis1977).Some
examplesare:askingrespondentstoindicatehowoftenonaveragetheyuseamotorcycle
(or car) when commuting (Chen & Chao 2010) or asking how often they performed each
behavioronaLikertscalerangingfromnevertoveryfrequently(Danner,Aarts&deVries
2008).

ResponseFrequencyMeasure(RFM)(Verplanken1994)isdifferenttofrequencyofbehavior
andshouldnotbeconfusedwithit.RFMworksundertheassumptionthatthechoiceofa
response is determined by a habitual choice in specific circumstances. Based on that
assumption,whenhabitisassessedusingRFM,participantshavetoprovideafastresponse
onwhatwouldtheychooseinthegivensituations.Ifarespondentanswersthathewould
usehiscartovisitcertainplaceinthisway,itispresumedusinghiscarwouldbehishabitin
suchsituation.

TheSelfReportHabitIndex(SRHI)isa12itemindexofhabitstrengthoriginallydeveloped
byVerplankenetal.(2003).SRHIisbasedonelementssuchas(1)historyofrepetition,(2)
automaticityunderstood as a lack of control, awareness and efficiencyand (3)
expressionofidentity.SRHIcorrelatedhighlywithFrequencyofBehaviorandtheResponse
FrequencyMeasure.Theindexhasshownreliabilityandvalidityinstudiesrelatedtoeating,
talking,transportation,andleisureandmentalhabits(Verplanken&Melkevik2008).

49

InarticlespublishedinjournalsofBusinessandEconomics,habithasbeenmeasuredusing
theSelfReportHabitIndex(SRHI)(Loibl,Kraybill&DeMay2011)and(Marchal2010),using
ResponseFrequencyMeasure(RFM)(Chen&Lai2011)and(Chen&Chao2010),andinthe
caseofHodgson(2010)habitwasnotmeasured.ForEducation,habithasbeenmeasured
using the SelfReport Habit Index (SRHI) (Kremers, van der Horst & Brug 2007). Articles
publishedinHealthandmedicine,showthathabithasbeenmeasuredusingtheSelfReport
HabitIndex(SRHI)(DeBruijn&Rhodes2010;DeBruijnetal.2007;deBruijnetal.2009;De
Bruijn&VanDenPutte2009;Kremers,vanderHorst&Brug2007).Gardneretal.(2011)
presentedacriticismoftheSelfReportHabitIndex,butdidtestit.

InPsychology,habithasbeenmeasuredusingtheSelfReportHabitIndex(SRHI)(DeBruijn
& Van Den Putte 2009; Lally et al. 2010; Loibl, Kraybill & DeMay 2011; Norman 2011;
Verplanken&Melkevik2008).TheResponseFrequencyMeasure(RFM)(Chen&Lai2011;
Klckner & Matthies 2004; Klckner, Matthies & Hunecke 2003) based on (Verplanken
1994).Abehaviorrecognitiontaskhasalsobeenusedtomeasurethecognitiveaccessibility
of participants' habitual and nonhabitual behaviors (Danner et al. 2011). Pahnila and
Warsta(2010)usedascaledevelopedbyLimayemandHirts(2003)inInformationSystems
which attempts to measure automaticity using elements of perceived addiction, habit,
natural use, unreflected use and use compulsion. Simple frequency of past behavior was
also used (Chen & Chao 2010) and Ouellette(1998) conducted a metaanalysis on studies
frequency of past behavior. In some cases, habit was not measured but theoretically
reviewedonly(Gardneretal.2011;Wood&Neal2007,2009).

ForInformationSystems,habithasbeenmeasuredusingtheSelfReportHabitIndex(SRHI)
(Pahnila,Siponen&Zheng2011;Polites2005).ThescaledevelopedbyLimayemandHirts
(2003) to measure habit as automaticity (Lankton, Wilson & Mao 2010; Limayem, Hirt &
Cheung2007;Pahnila&Warsta2010).Polites(2009)developedascaleforperceivedhabit
whererespondentsreportontowhatextenttheusagebehaviorisdrivenbyhabit.Polites
(2009) included some items from Limayem et al. (2007) on utilization and choice,
naturalness and automaticity. Liao et al. (2006) used a scale adapted from (Gefen 2003)
which considers usual behavior, preference, first choice and perceived frequency of

50

behavior.Inonecasewherehabitwasmeasured(Guetal.2010)thescaleisnotprovided,it
was noted that such study was conducted in collaboration with a private company
(Microsoft).InthecaseofGuineaandMarkus(2009)atheoreticalreviewisprovided,but
nomeasurementisaccompanied.

2.11.1 Measuringtheotherhabits:agapintheliterature

Fromtheprototypedefinitionofhabitshownbefore(Section2.7Prototypedefinitionof
habits),itisimportanttonoticethatliteratureofhabitsuggeststhathabitsdonotexistina
pureandisolatedform.ConsistentwithWozniak(2009),Swartz(2002),andBourdieu(1984)
habits are called systems and structures, which indicates relationships of interaction and
support among habits. However, it is rare finding direct mentions to this matter in the
literatureofhabitandbehavior.

WhenAjzen(2002)orGardner(2011)etal.wroteAmoreappropriateapproachwouldrely
on an operationalization of habit that is independent of the behavior it is supposed to
explainandpredict(Ajzen2002,p.14;Gardneretal.2011,p.141),theynecessarilyimply
that there are other habits that explain and predict behavioral intention and actual
behavior. Empirical evidence of the other habits is scarce in literature of habit and
behavior.However,itcanbefoundinstudies measuringmultipleandnonpredetermined
habits(Table2.1GapinLiterature).

ThecriteriausedtoclassifythemeasurementcasesinthearticlesrepresentedinTable2.1
are the following. Measures are considered single habit when only one habitrelated
behavior is considered in the final indicator or score. Multiple habit measures are those
whichconsidertwoormorehabitrelatedbehaviorstoberepresentedinonefinalindicator
or score. The measurements for habit(s) were grouped as: predetermined, when the
researcher decided and explicitly stated the habitrelated behavior; semipredetermined,
when the researcher set a general constraint or rule but did not specify the habitrelated
behavior; and nonpredetermined, when the respondent was free to introduce any habit
relatedbehaviorfromanydomain.

51

Mostofthemeasurementcases(51/60)ofhabitandbehaviorhavefocusedonmeasuringa
single predetermined habit. This is, only one habit is measured at the time and the
researcherdeterminedit.Themeasuredhabitisusuallythesameastheproposedbehavior.
Out of 51 cases of measurement of single predetermined habit, 21 were related to using
Information and Communication Technologies (ICT), 8 to health, 9 to food or drinks
consumption, 11 to transport, 2 to doing exercise, 6 to others themes. The most popular
analysismethodwasregressionusedin28ofthecases,PartialLeastSquares14,SEM14,
and1other.Theaveragesamplesizeforsinglepredeterminedhabitwas565(fordetailssee
Appendix5).

2.11.2 MeasuresofMultiplePredeterminedHabits

Four studies found in this review were measuring multiplepredetermined (MP) habits at
the same time. Landis, Triandis and Adamopoulos (1978) conducted a research on
classroom teacher behavior with 77 school teachers. To capture habits the study used
frequency of the observed behavior. In intervals of ten seconds an observer recorded all
relevantteachersbehaviorsinpredeterminedcategories.Thefrequencyoftheactionswas
usedtodeterminehabitstrengthofvariousbehaviorsatthesametime.

Verplanken (1994) introduced responsefrequency measure (RFM). Verplanken tested a

modeloftravelmodechoicewithdatacollectedamong199adultsfromavillage.Habitwas
measuredusingtenimaginarysituationsthatrequiredachoiceoftravelmode.Thetarget
habitrelatedbehaviorwascarutilization.Habitwasthenacompositevariablethatranged
from0to10.Itsvaluedependedonhowmanytimescarwasselectedfromthesixoptions
available(bicycle,bus,cab,car,trainandwalking).

VerplankenandFaes(1999)conductedafieldexperimentaboutunhealthyfoodhabitswith
102studentsasrespondents.ThestudyusedtheTheoryofPlannedBehaviorasframework,
andavariationoftheresponsefrequencymeasure(RFM)toassesshabits.Aquestionnaire
presentedalistwith67differentfoods.Participantscheckedproductstheyconsumedinthe
last week. Unknown to the respondent, 37 of the options in the list were counted as
unhealthy. The number of unhealthy foods selected by the respondent was used as a

52

measure of unhealthy habits. All the foods were determined by the researcher, and the
variousselectionswereaggregatedinonesinglescore.

Klckner,MatthiesandHuneckes(2003)studypresentsonemultiplepredetermined(MP)
measure for habits in a bipolar measure (0=car/1=subway). However, their study contains
measuresthatfallintofourofthesixcategoriesinthisthesis.Inthereport,itispossibleto
find singlepredetermined, multiplepredetermined, singlesemipredetermined and
multiplesemipredeterminedmeasurementsforhabit(SP,MP,SS&MS).Thepurposes
ofKlckner,MatthiesandHuneckes(2003)studyweretointegratehabitintotheprocess
of normative decision making to predict behavior, and improve the operationalization of
habit.Thetargetbehavioristheuseofcarasmeansoftransport.Thestudywasconducted
inGermanywith160participants,anditwasanalyzedwithmultipleregressions.

Chen and Lai (2011) studied push strategies to reduce the usage demand of motorized
vehicles,andpullstrategiestoattractmorepublictransportusers.BasedontheTheoryof
Planned Behavior and using responsefrequency measure (RFM), 231 commuters were
surveyedinTaipeiandKaohsiung.Sevenstatementsrepresentedimaginarysituationsthat
requiredtraveling.Therespondenthadthreeoptionsineverycase:usingmotorcycle,caror
public transport. The strength of each habit was assessed by the number of times each
habitrelatedbehaviorwasselected.

2.11.3 Measuresforsingleandmultiplesemipredeterminedhabit

Klckner, Matthies and Huneckes (2003) study, previously mentioned in Section 2.11.2 ,
also offers measures for single and multiple semipredetermined. These measures can be
found under the label of Original RFM and the Multple RFM in (Klckner, Matthies &
Hunecke 2003). Although the authors report only a slight modification to Verplankens
(1994) RFM, conceptually, this modification is not minor. While Verplanken provided a
specific listof six options for the respondents, Klckner,Matthies and Hunecke made this
itemintoanopenquestion.Duringaninterviewtherespondentsweremeanttoanswerthe
firstmodeoftravelthatcametotheirmind.Eventhough,therewasarestrictionsetbythe
researcher,therespondentshadthechancetocomeupwithnonpredeterminedoptions.

53

Inthesamestudy,allthechosentravelselectionswereusedtodetermineavalueforthe
Multple RFM. In the multiple RFM, the researcher calculated the ratio including the
numberofcarmentionsdividedbythenumberofanyotherchoicesmadeinthesameitem.

Inbothcases,OriginalRFMandtheMultpleRFM,thevariabledependedonthecontent
of an open question. However, these measurements were not classified as non
predetermined because the respondents were not free to mention any habit or behavior
thatcametotheirminds.Theywererestrictedtoadomainsuchastravelmodeschoices.
The possible answers are limited, and then can be considered under a high level of the
researchers control. Thus, these types of measurements were categorized as semi
predetermined.

Despite the reported failure to improve RFM, Klckner, Matthies and Huneckes (2003)
methodisquiteuniqueinthatitoffersoneofthemostinterestingapproachestomeasure
habits. Other modification made to the RFM could not overcome the Original RFM. The
reportconcludedthattheOriginalRFMshouldremainunmodified.However,theOriginal
RFMwasindeedamodifiedRFM,differenttotheoriginalinthelevelofpredetermination
ofthehabitrelatedbehavior.

2.11.4 Measuresforsingleandmultiplenonpredeterminedhabits

An interesting example of how nonpredetermined habits can be captured is present in

VerplankenandOrbells(2003)work.Theirresearchreportcontainsfourstudies.Thefour
studiesareontravelmodechoice.Study1,2and3capturehabitwiththeSelfReportHabit
Index (SRHI), a measure of single predetermined habit. However, Study 4 reveals an
uncommonapproachtomeasuringhabits.

Seventysix undergraduate students participated in a laboratory experiment in The

Netherlands. In the first of two sessions, participants listed their own daily and weekly
habits.Then,theyassignedafrequencyofperformancetoeachhabit.Inthesecondsession,
the researchers selected one habit for each participant based on the selfreport on
performancefrequency.TheselectedhabitwasplacedinaSRHI11pointscale.Eventhough

54

theselectionofthehabitpartiallydependontheresearcherscriteria(highestselfreported
frequency), it greatly depends on the respondents selection of their habits and the
frequency they reported. This is why this measurement was classified as non
predetermined.

Verplanken and Orbell (2003) studies tested only the reliability and validity of the SRHI
scale. Study 4 revealed that daily habits get a higher SRHI than weekly habits. However,
behavioralintentionoractualbehaviorswerenotconsidered.Althoughthisreviewfocuses
on studies testing habit and intention or habit and behavior, SRHI is one of the most
commonly used scales in habit and behavior. Besides, Verplanken and Orbells work is
uniqueinthatitincludesoneofthefirstmeasurementsofnonpredeterminedhabits.

Of the currently available measurements of habit, RFM has been demonstrated as one of
the most versatile measurements. As shown in the previous paragraphs, it has been the
basisofmeasuringotherthansinglepredeterminedhabits.

The literature review conducted for this thesis did not find any scale or measurement
attemptingtocapturemultiplenonpredeterminedhabits.Thismayconstituteanimportant
gap in the literature of habit, information systems, and behavior. If habits do not exist in
isolated and pure forms, if they inform our taste, choices, and understanding (Bourdieu
1984; Swartz 2002; Wozniak 2009), apparently unrelated habits might be supporting or
interferingwithnewbehavioralpropositions.

A travel mode choice (using a car) could be influenced by the habit of wearing a type of
shoes (for example high heels) to the same extent it might be influenced by the habit of
usingotherkindsoftransportation(forexamplebus,train,bicycle).Thechoiceofusingan
Information and Communication Technology (ICT) might also get support from habits,
relatedtothewaysindividualswork,spendleisuretime,dress,play,etc.,andnotonlyfrom
theirhabitsofusingthetechnologyitself.

Thissectionpresentedempiricalevidenceoftherelationshipofhabit,behavioralintention
and actual behavior. It established that the most consistent and reliable relationship was

55

habit to behavioral intention. This hypothesis failed to be supported in only 3.6% of the
studies,whereastherelationshiphabittoactualbehaviorfailedtobesupportedin13.9%of
thecases.Thissectionalsodiscussedthemostsalientmeasurementsofhabit(frequencyof
behavior,theResponseFrequencyMeasureRFM,andtheSelfReportHabitIndexSRHI).
Since literature suggests that habits do not exist in a pure and isolated form, this review
points to a gap in the measurement of multiple nonpredetermined habits. This thesis
intendstoaddressthisgap.

Section2.12explorestheconceptoffitfromtheperspectiveofPersonEnvironment[PE]in
psychology.Theconceptoffitisusedtodevelopameasurementforhabitsabletocapture
moreofthoseotherhabitswhichareunrelatedtothetargetbehavior.

2.12 Personenvironmentfit

Literature on habit and PersonEnvironment [PE] fit informed the development of the
habittechnology fit (HTF) construct, and measurement capable of capturing more than
single or predetermined habits. This section reviews the different classes of fit in the
literatureandtheappropriatenessofperceivedfittocapturetheotherhabits.

Theory on PersonEnvironment [PE] is considered one of the most respected lines of

psychological theorizing (Dawis 1992; KristofBrown & Billsberry 2012; KristofBrown,
Zimmerman&Johnson2005).Mostoftheattentioninresearchinregardtofithasbeen
placed upon the person and their vocation, job, organization, workgroup, situation, and
supervision(seeEhrhart&Makransky2007;Horveraketal.2013;HyeHyun2013;Kim&Kim
2013;Marcus&Wagner2013).Besides,avarietyofdimensionshavebeenmeasured.These
includeskills,needs,preferences,values,personalitytraits,goals,andattitudes.

Fromaperspective,itisreasonabletoconsiderhabitsaspersonalcharacteristicsinorderto
answerthecallsforresearchonthearea.Abelson(Abelson1981;Verplanken,Myrbakk&
Rudi2005)suggestedhabitsareascriptofapersonalkind,Lanktonetal.(2010)recognizes
habitsasselfidentifying,andNorman(Norman2011)makesreferencetohabitsreflectinga
persons sense of identity. For Danner habits are the result of ones personal history of

56

behavior, making habits unique, and idiosyncratic (Danner et al. 2011, p. 3). Therefore,
research on habits might acceptably be considered a relevant subset of personal
characteristics.

Thus,thisthesishasassumedhabitscanbeconsideredapartofapersonscharacteristics,
andtechnologyacomponentofhisorherenvironment.Theoryonpersonenvironment[P
E]hasbeenextrapolatedtopersonvocation,personjob,personorganization,andperson
situation. This may suggest that PE fit could also be extrapolated to the study of habits
(person)andtechnology(environment).

Fit has been defined as the match of two related variables. A measure of fit is developed
independently of any dependent variable. Fit as a match can be in a range of 0 to 1 (0%
100%), and perfect fit occurs when the planed system matches a required ideal system
(Venkatraman&Camillus1984).Fithasalsobeenexplainedassimilarity,needsatisfaction,
anddemandabilitymatch(KristofBrown,Zimmerman&Johnson2005).

However,thisthesisborrowsandadaptsKristofBrown,ZimmermanandJohnsons(2005)
definitioninwhichpersonenvironmentfitisthecompatibilitybetweenanindividualanda
work environment that occurs when their characteristics are well matched (p. 281). This
definitionisgivenforperceivedfitandemphasizesintheaspectofcompatibilityintheway
KristofBrownmakesadistinctionfromsubjectiveandobjectivefit(Niessen,Swarowsky&
Leiz2010).

Analogous to the ways it has been used, according to KristofBrown, Zimmerman and
Johnson (2012; 2005), only perceived fit would be suitable to address the problem of
measuring multiple nonpredetermined habits. Perceived fit has dynamically captured
characteristics which are salient to the respondents. It allows them to define the level of
importance and compatibility between the aspects of the environment and those of their
own person. Perceivedfit does not require more than one source of information and it is
consideredaholisticassessmentwhichismorepronetoconsistencyeffects(KristofBrown,
Zimmerman&Johnson2005).

57

Implicitly,byselectingperceivedfitasameasurementapproach,theresearcherwouldbe
measuring relative fit, either supplementary or complementary fit, fit as a profile
deviation',andeither'physicalorcognitivefit'(Avital&Te'eni2009;Buxton1986;Carless
2005;KristofBrown,Zimmerman&Johnson2005;Kristof1996;vanVianen,DePater&Van
Dijk2007;Venkatraman&Camillus1984).

2.13 Classificationsoffitinliterature

Aninterpretationoftheperspectivesthatclassifytheconceptoffitispresentednext.This
thesis presents the diverse types of fit defined by five sorting perspectives. In these, fit is
definedbyapointofview,itslevelofspecificity,uniontypeitrepresents,levelofbelonging,
andlevelofobservability.

2.13.1 Fitdefinedbypointofview(perceived,subjectiveorobjectivefit)

In evaluating the level of match, three main classes of fit can be identified: Perceived,
SubjectiveandObjectiveFit(KristofBrown,Zimmerman&Johnson2005).Perceivedfitisa
directassessmentofmatch,reportedbyarespondentinasinglevariable;Subjectivefitis
thelevelofmatchoftwovariablesduringtheanalysis,reportedbyarespondentseparately;
andfinallyObjectivefitisthedegreeofmatchbetweentwovariablesduringtheanalysis,
but in contrast with perceived and subjective fit, the values of the variables do not come
fromanindividualwhosefitisunderassessment.Instead,inobjectivefit,dataiscollected
fromothersources.

2.13.2 Fitdefinedbylevelofspecificity(absoluteorrelativefit)

VanVianenetal.(2007)refertoAbsoluteorRelativeFit,whichcanbeclassifiedbythelevel
of specificity in the variables they comprise. While Absolute fit is a match evaluation
involving mathematical calculation of discrepancy, Relative fit is a personal rating
evaluating fit. Subjective and Objective fit as defined by (KristofBrown & Billsberry 2012;
KristofBrown,Zimmerman&Johnson2005)canbeclassifiedasabsolutefitasdefinedby
Van Vianen, as well as Perceived fit can be put together with Relative fit. Absolut fit

58

requiresdetailedspecificityintheaspectsunderassessment,whileRelativefitisdescribeda
holisticapproachwhichcaninvolvethoseaspectswhicharerelevanttotherespondent.

2.13.3 Fitdefinedbyuniontype(complementaryandsupplementaryfit)

A fit or compatibility has two classes in which a union between two entities (person and
environment)mayfallwhentheyareputtogether.Theirrelationcanbecomplementaryor
supplementaryintermsoffit.Ononehand,Supplementaryfitoccurswhenanentity(such
asPerson[P])hassimilarcharacteristicstootherentity(anaspectoftheEnvironment[E]).
OntheotherhandComplementaryfittakeplacewhentheperson[P]andthesituation[E]
meeteachother'sneeds(Carless2005;Kristof1996;Muchinsky&Monahan1987).

2.13.4 Fitdefinedbythelevelofbelonging(FitasaGestaltorasaprofiledeviation)

Venkatraman and Camillus (1984) identified two classes of fit which can be defined by
belonging. The first one is fit as a Gestalt, which is defined as the degree of internal
coherence among a set of attributes, i.e., how well certain entities belong to clusters (or
taxonomies) depending on their characteristic combination of values for the relevant
variablesaccountedinacase.Thesecondoneisfitasaprofiledeviationisthedegreeof
adherencetoaspecificprofilewhichisspecifiedexternallyandisanchoredtoadependent
variable. For example, the level in which a technology complies with the final consumers
requirementderivinginapurchase,orinversely,thelevelofskillofanemployeeinorder
tooperateamachinewhichmightderiveinjobperformance.

2.13.5 Fitdefinedbythelevelofobservability(physicalorcognitivefit)

Physicalfitreferstoagoodadaptationbetweenshapes,materials,mechanismsanddesigns,
ononeside,totherequirementsoftheother(forexampleinergonomicsadesignwhich
allows comfortable operation and minimal physical effort) (Avital & Te'eni 2009; Buxton
1986). Cognitive fit would occur when the information mental models within a person
matches with the task demands (Avital & Te'eni 2009; van Vianen, De Pater & Van Dijk
2007).

59

2.14 Theoreticalrelationshipsoffit

2.14.1 BehavioralIntentionandBehavior

Intherelationshipwithbehaviorandbehavioralintention,perceivedandsubjectivefithave
been contrasted with objective fit. The first two have been found better determinants of
behavior(Cable&DeRue2002;KristofBrown,Zimmerman&Johnson2005).Annelies,De
PaterandFloorVan(2007)supportthisarguingthatsamesourcemeasuresoffitseemto
havestrongerrelationshipswithindividualoutcomes.Inthecontextofindustrialrelations
and human resources, individual outcomes refer particularly to turnover intention which
mightbeconsideredaformofbehavioralintention(Annelies,DePater&FloorVan2007).

Furthermore, in the same context of PersonEnvironment [PE] studies, Cable and DeRue
(2002), Cable and Judge (1996), Newton and Jimmieson (2008), and Verquer(2002) agree
thatlowsubjectivefitisconnectedwithhigherturnoverintentions(behavioralintentions).

Caplan(1987)andCarless(2005)affirmthatitisindividualfitperceptionsandnotobjective
fitthatimpactsifanindividualpursuesworkwithanorganization.Consequently,individual
perceptions of fit are posited as more accurate representations of personal reality than
objectivefit.

Chatterjee(2010)presentedempiricalevidenceontherelationshipofperceptionsoffite
servicesadoption.Thestudyfoundapositivecorrelation,anditalsofoundthattheeffect
was stronger for current users than for prospective users. This suggests that experience,
understood as the length of time a person has used a technology, may be posed as a
moderatorinthisrelationship.

Thisthesisconsidersthetheoreticalcharacteristicsofhabitandfit.Itconsidersevidenceof
thepositiverelationshipbetweenhabitandbehavioralintention,andtheempiricalresearch
that suggest a positive relationship between perceived fit and behavioral intention.
Supported on those the relationship of habittechnology fit and behavioral intention is
hypothesizedtobepositiveandsignificant.

60

2.14.2 Otherrelationshipsoffit

Literaturehassuggestedothersignificantrelationshipsforfit.Inregardstoemployeejob
related attitudes, research has provided empirical evidence that subjective/perceived fit
may positively predict job satisfaction (Cable & Judge 1996; Newton & Jimmieson 2008;
Verquer 2002). Comparing objective with subjective/perceived fit, evidence reveals that
individuals perceptions are better predictors of attitudes than objective measures of fit
(Carless2005;Judge&Cable1997).

Thus there is strong evidence about the impact of fit perceptions upon attitudes and
behaviors, for instance satisfaction, commitment, and job turnover (Cable & DeRue 2002;
KristofBrown,Zimmerman&Johnson2005;Niessen,Swarowsky&Leiz2010).

Besides, Mullany, Tan and Gallupe (2007) argue that information and communication
technologies and their interfaces should be designed to fit the cognitive style of the user.
Otherwise,thecognitivestylegapbetweentheuserandcreatorwhichisimprintedinthe
technologywouldreduceusersatisfaction.

2.15 Potentialadequacytomeasurehabits

The concept of PersonEnvironment [PE] (KristofBrown, Zimmerman & Johnson 2005)

perceived fit was borrowed from Personnel Psychology literature because of its potential
adequacytomeasuringhabits.Perceivedfit(oftenusedindistinctlywithsubjectivefit)have
resultedbetterdeterminantsofattitudes,intentionsandbehaviorthanobjectivefit(actual
fit) (Cable & DeRue 2002). It has been successfully used matching diverse aspects of a
person and their environment, in similar ways (see Karahanna, Agarwal & Angst 2006;
Resick,Baltes&WalkerShantz2007;Saks&Ashforth2002)tohowthehabittechnologyfit
constructisconceivedinthisresearch.

The method to measure perceived fit is straight forward (Annelies, De Pater & Floor Van
2007).Respondentshavetobeaskedtoestimatethecongruencebetweenthepersonand
the environment. The following is an example of a measurement item, appropriate to be
used in a rating scale for the assessment of compatibility between personorganization
61

mypersonalvaluesmatchmyorganizations'valuesandculture(Annelies,DePater&Floor
Van2007,p.191).

Perceived fit is a direct assessment reported by individuals, allowing them complete

cognitive manipulation of the evaluation. However, this manipulation permits applying a
personalweightingschemetothecountlessaspectsofthepersonandhisenvironment.This
facilitates individual differences in importance or salience of various dimensions to be
capturedinasinglescore(KristofBrown,Zimmerman&Johnson2005;Kristof1996).Habit
is a multidimensional construct and its behavioral instances innumerable. Therefore,
capturingallinterveninghabitswhichareapplicabletoasituationmightbedifficult,ifnot
impossible through approaches to measure fit, other than Perceivedfit. Thus, capturing
selfselected habits weighted by their relevance and their fit to a behavioral proposal of
utilizationjustifiesthisapproach.

This approach may also serve as a proxy to the concept of habit. Section 2.7 suggests a
definition based on the essential characteristics of habit as found in the literature. Such
definitionsuggestshighdimensionalityandcomplexityfortheconceptofhabit.Different
disciplines have paid attention to different dimensions of habit, such as automaticity,
frequency or past behavior, see (Verplanken 1994), (Wood, Tam & Witt 2005), and
(Thgersen 2006). Perceived fit as a latent variable, allows individuals (respondents)
complete cognitive manipulation of their evaluation. Individuals are allowed to define the
salience of the various dimensions of the variable to be apprehended in their answer
(KristofBrown, Zimmerman & Johnson 2005; Kristof 1996). This suggests a property of
perceivedfit,whichitmaybeabletocapturethehabitswhicharemostimportanttothe
individual, but also serve to capture the salient dimensions of habit from the individuals
perspective.

Aclearandusefuldistinctionbetweentheformertypeoffitversussubjectiveandobjective
fit is offered by Kristof (KristofBrown, Zimmerman & Johnson 2005; 1996) (see Section
2.13.1 ) to conceptually guide research and measurement of fit. In subjective fit, the
respondentreportsvariablesdefinedfortheperson[P]andfortheenvironment[E].Both
[P] and [E] are assessed indirectly through comparison during the analysis. Objective fit
62

measures and evaluations of [P] and [E] are done separately and indirectly. However, the
sourcesofthereport,unlikeinperceivedandsubjectivefit,areotherssourcesratherthan
theindividualforwhomfitisunderassessment.

HabitTechnologyFitconstructtakesanindividualcharacteristicoftheperson[P]habit
and a complex aspect of the environment [E]the behavioral proposition of using a
suggestedtechnologyinasimilarapproachaspreviousresearchon[PE]fit.Examplescan
be found for constructs such as: person[P]organization[E], abilities[P]demands[E],
needs[P]supplies[E] (Resick, Baltes & Walker Shantz 2007); workstyle[P], existing work
practices[P], prior experience[P], and values[P] in the use of technology [E] (Karahanna,
Agarwal&Angst2006);person[P]job[E],person[P]organization[E](Saks&Ashforth2002).
Inthesamewaythisstudytakeshabit[P]infitwithanexternalproposalofbehaviorthe
usageofasuggestedtechnology[E].

2.16 Potentiallimitationsofmeasuringhabitswithperceivedfit

In the previous section, this review has presented perceived fit as a potentially adequate
concept that may allow capturing salient characteristics of the person and their
environment.Perceivedhabithasbeenselectedinthisthesistoapproachthemeasurement
of multiple nonpredetermined habits. However, this approach comes with limitations
concerningitsmeasurementandcapacitytoprovideadditionaldeeperinformation.

Thefirstlimitationofperceivedfitisameasuringissue.Itconfoundstheconstructsofthe
person and environment. This implies that it is not possible to estimate the independent
effectsofthepersonscharacteristicsapartfromtheenvironments(Ahmad2011;Edwards
&Shipp2007;Edwards1991,1996).

Some examples of perceived technology (environment) characteristics, can be constructs

such as perceived ease of use, and perceived usefulness (Davis 1989) in TAM, or effort
expectancyandperformanceexpectancy(Venkateshetal.2003)inUTAUT.Thesevariables
only reflect perception about characteristics of technology in its measurement. They
excludepersonalcharacteristics;stilltheyarerelativetotheperson.Theydependonwhat

63

the respondents want to do, and what their ability is. However, compatibility and
usefulness,twoconceptuallydifferentfactors,highlycrossloadoneachother(Karahanna,
Agarwal & Angst 2006; Karahanna, Straub & Chervany 1999; Moore & Benbasat 1991).
Karahanna (2006) conclusion is that people might not see technology as useful if it is not
compatiblewiththeirworkstyle,whichsuggestapossibleconsequenceoftheconfounding
effectfromperceivedfit.

Asecondlimitationisthatperceivedfitisalsorestrictedinregardstotheinformationitcan
provide. Because most of the assessment of fit occurs in the head of the person, the
outcome does not offer the details about aspects the respondent considered as salient
(Annelies,DePater&FloorVan2007;Cable,D.M.&DeRue,D.S.2002).VanVianenetal.
(van Vianen, De Pater & Van Dijk 2007) argues that while perceived fit is able to capture
salient and unique combination of aspects chosen by the individual, it does not provide
understandingonwhichspecificaspectsarecausingmisfitorcognitivedissonance.

A third limitation is in regards to the direction of some compatibility relationships. Some

variables of the environment like moral values may result incompatible with the person.
This is because the value standards in the environment can be too high or too low in the
persons opinion. Perceived fit general measures cannot provide information on the
directionofthemisfitVanVianenetal.(2007).Thislimitationmightnotbeapplicableinthe
caseofmeasuringhabit,ashabitcannotbetoohighortoolow.

2.17 TechnologyFitandBehavior

Intheliteratureofinformationsystems,theconceptoffithasbeenusedbefore.Mostof
the interest has fallen upon the match of technology and the task to be performed. Less
attentionhasbeengiventopersonalcharacteristics.However,recentworkhasaddressed
severaldimensionsofcompatibilityinrelationshipwithbehavior,see(Karahanna,Agarwal
&Angst2006).

64

2.17.1 TechnologytoPerformanceChain

TechnologytoPerformanceChain(TPC)isamodelwithfocusonTaskTechnologyFit(TTF)
and individual performance. TPC has its roots in the work of DeLone and McLean (1992)
which posits that utilization and user attitudes about technology lead to performance
impacts. However, TPC highlights the role of TaskTechnology Fit in explaining how to
achieve individual performance. The basic beliefs of this theory for a positive impact in
performanceare:technologymusttobeutilizedandmustshowagoodfitwiththetasksto
execute(Goodhue&Thompson1995).

Goodhue & Thompsons (1995) model suggests that Task Characteristics, Technology
CharacteristicsandIndividualCharacteristicshaveapositiveimpactuponTaskTechnology
Fit.TaskTechnologyFitthenhasanimpactuponUtilizationandPerformance.

Although, the relation between TTF and utilization showed little support in the original
study (Goodhue & Thompson 1995), later studies found TTF a significant determinant of
utilization. (Lee, Lee & Kim 2005) studied a sample of 110 users in the context of mobile
commerce (Mcommerce). The findings indicate that the task, technology, and individual
user characteristics positively affect TaskTechnology Fit and Mcommerce usage. (Lin &
Huang2009)studiedasampleof194EKR(ElectronicKnowledgeRepositories)users.Task
technologyfittheorywassupportedasakeyfactorindeterminingtheEKRusage.Yenetal.
(Yen et al. 2010) did a study on mobile commerce adoption. Results showed that Task
TechnologyFitisasignificantdirectpredictoroftechnologyadoptionintention.

Recently, a different result emerged. McGill et al. (2011) studied TTF and adoption of a
Learning Management System. His results showed that TTF in not associated in a simple
linearwaywithutilization,wherepoorfitcanbeassociatedwithlowutilization.

This model was able to explain 38% of the variance of TaskTechnology Fit and 41% of
utilization. According to the comparison made in the study the new model outperformed
TAMandTTFalone(Dishaw&Strong1999).

65

2.17.2 FitandAppropriationModel

Dennis, Wixom & Vandenbergs (2001) Fit and Appropriation model (FAM) found
inconsistent findings in the past GSS (Group Support Systems) research, and considered
important to develop a new model to explain the fit and appropriation of such kind of
informationsystem.

ThemodelpresentedbyDennisetal.(2001)suggeststhatAppropriationSupport,andthe
fit between [System] Capabilities and Task have an impact upon Appropriation. Habitual
Routines were identified as an important determinant of Appropriation, but were not
measured, as the study consisted in a metaanalysis that required past research to have
measuredit,andsuchmeasurementwasnotavailable.Theauthorsconcludedthatwhen
usingthistheoreticallens,theresultsofGSSresearchdonotappearinconsistent(Dennis,
Wixom&Vandenberg2001,p.167).

Numerousauthorshaveidentifiedandmeasuredtechnologyfitpayingspecialattentionto
tasktechnology fit. Literature identifies some relevant determinants of technology fit:
individualcharacteristics;taskcharacteristics;technologycharacteristicsandtheprecursors
of utilization (Goodhue & Thompson 1995); task requirements and tool functionality
(Dishaw & Strong 1999); capabilities and task (Dennis, Wixom & Vandenberg 2001); and
humandrives(Junglas,Abraham&Ives2009).Althoughdeterminantsvaryamongauthors,
moststudiesagreethatTechnologyFithasapositiveimpactuponadoptionofatechnology
(Dennis,Wixom&Vandenberg2001;Dishaw&Strong1999;Junglas,Abraham&Ives2009).

Technology fit provides minimal direct support to the conceptualization of the habit
technologyfitconstruct.Still,itisconsistentwiththeideathatthebetterthefit,thehigher
the acceptance. Technology research has concentrated on primarily on the fit of task and
technology,whichprovidesunderstandingonpreviousperspectivesinthefieldandpossible
gaps,suchasincludinghabittechnologyfitinthislineofresearch.

66

2.18 Definitionofhabittechnologyfit

Perceived fit has been proposed as an adequate approach to study a wider structure of
multiple nonpredetermined habits (see Section 2.15 Potential adequacy to measure
habits),but alsotocoverabroadernumberofthedimensionsofhabit (seeSection2.7).
Perceived fit provides an evaluation of compatibility between person (habits) and
environment(technology)characteristics(Cable&DeRue2002;KristofBrown,Zimmerman
&Johnson2005).Theseideasalongthechapterhavebeenbuildingtheconceptualizationof
anewconstruct:Habittechnologyfit.Hereisanoperationaldefinitionofthisconcept:

Habittechnology fit is defined in this thesis as the degree to which an individual believes
thatusingthetechnologyiscompatiblewithhisorherhabits.

2.19 Technologyacceptance

This thesis has discussed habit as the main focus of this research, perceived fit as an
approach to capture a wider number of habits and a broader number of dimensions of
habits. This section introduces a third and last component, the theoretical context of
interestofthisstudy,technologyacceptance.

The history of technology acceptance has seen several models describing theoretical
relationsbetweenconstructs.Someofthosemodelshaveputtogethertheorythatmatures
throughempiricalvalidation.Someofthosetheorieshavebecomeseminalanddominated
thefieldoftechnologyacceptance;seeTRA(Fishbein&Ajzen1975),TPB(Ajzen1991),and
TAM (Davis 1989). Other researchers have used these models, extending and combining
them.Stillonlyafewhaveremainedthepillarsoftechnologyacceptanceresearch.Therest
ofthemodelsworkwithvariablesaroundthebasicconstructs(Cornacchia,Baroncini&Livi
2008).

These basic constructs around behavioral intention and actual behavior have provided
foundation to the technology acceptance research, and have maintained their focus into
explaining the characteristic underlying concept (Venkatesh et al. 2003), shown in Figure
2.3.Theunderlyingconceptoftechnologyacceptanceresearchbeginswithindividualsbeing
67

exposed to a technology. They react to it, and form an intention. Finally, intention
determinesactualbehavior.Explainingthecausesofintentionandactualbehaviorhasbeen
theessenceunderlyingthisfield.

CORETHEORYOFTECHNOLOGYACCEPTANCE

Figure2.3CoreTheoryofTechnologyAcceptance (Venkateshetal.2003)

However,theunderlyingconceptanditsbasicrelationshipsarenotexclusiveoftechnology
acceptance.Someofthosetheoreticalrelationshipswereinheritedfromlongtraditionsof
researchofbehaviorinsocialpsychology,forexampletheTheoryofReasonedAction(Ajzen
&Fishbein1980;Fishbein&Ajzen1975).Theserelationshipswereofgeneralapplicationto
studybehaviors.ItwasperhapsuntiltheTechnologyAcceptanceModel,TAM,(Davis1989)
thatamodelwastailoredspecificallyfortechnology(Venkateshetal.2003).

Technology acceptance research began to develop since the 1970s, when Information
Systemsresearchersbegunidentifyingtheconditionsthatmayfacilitatetheintegrationof
the emerging computers in business (Legris, Ingham & Collerette 2003; Teo, Lee & Chai
2008).Someofthemaindriversofthislineofresearchhavebeenthehighrateoffailureof
newtechnologies(Morris&Venkatesh2010;Pan,Hackney&Pan2008),themagnitudesof
capital being invested on new technology, and the increasing rate at which that capital is
beinginvested.Anotherunderlyingbeliefintechnologyacceptanceisthatfortechnologyto
yield the results, it first has to be accepted and used (Venkatesh et al. 2003; Westland &
Clark2001).Someoftheproblemsinacceptanceandusearetechnologybeingoversoldand
notmatchingtheexpectationsinreality;alsothefastpacedchangestowhichtechnologies
aresubject(Cornacchia,Baroncini&Livi2008).

68

This section introduced the essence of the theoretical context of habittechnology fit.
Technology acceptance is further discussed. However, the elements covered until now
sufficetoinformthefirstresearchquestionofthisthesis.

2.20 Researchquestion1

This thesis has uncovered a gap by reviewing the literature of habit (see Section 2.2 and
Table 2.1). Wider structures of habits have rarely been studied in relation to behavioral
intention, except for the habit that corresponds to the target behavior. Therefore, the
impactofthosemultiplehabitsuponbehavioralintentionremainsunknown.

It has been discussed that habits do not exist isolated but in structuring structures.
Furthermore, they inform individuals taste, choices, and understanding (Bourdieu 1984;
Swartz 2002; Wozniak 2009). Habits in relationship with behavior have been extensively
theorized (see Hodgson 2010), and single habits have been empirically confirmed as a
determinantofintentionandbehavior,forinstancein(EscobarRodrguez&CarvajalTrujillo
2013),(Huang,Wu&Chou2013)and(Venkatesh,Thong&Xu2012).

Perceived fit has been proposed as an adequate approach to study a wider structure of
multiple nonpredetermined habits (see Section 2.15 Potential adequacy to measure
habits),but alsotocoverabroadernumberofthedimensionsofhabit (seeSection2.7).
Perceived fit provides an evaluation of compatibility between person (habits) and
environment(technology)characteristics(Cable&DeRue2002;KristofBrown,Zimmerman
& Johnson 2005), which this thesis applies to technology acceptance in a newly
conceptualizedhabittechnologyfit construct.However,theimpactofhabittechnologyfit
uponbehavioralintentionisunknown.Thus,thefirstresearchquestionarises:

RQ1Whatistheimpactofhabittechnologyfituponbehavioralintentioninthe
contextoftechnologyacceptance?(H1&H1a)

Researchquestion1informsHypotheses1and1adevelopedinthefollowingsection.The
confirmationorrejectionofthesehypotheseswillprovideananswertothisfirstquestion.

69

2.21 Hypothesis1and1a

The first research question asks about the impact of habittechnology fit upon behavioral
intention.Themagnitudeanddirectionoftheimpactaredeterminedbytworelationships.
Thesearethedirectrelationshipbetweenhabittechnologyfitandbehavioralintention,and
theonewithitsmoderatingvariables(Whisman&McClelland2005).Therefore,toaddress
thefirstresearchquestionitwasdeemedappropriatetopositHypothesis1,whichrefersto
the direct relationship, and Hypothesis 1a, which evaluates the moderators of habit
technologyfitandbehavioralintention.

The conjectures behind Hypothesis 1 derive from previous theoretical and empirical work
related to the main components of habittechnology fit. These components are habit and
perceived fit. For habit, extensive empirical evidence of its relationship with behavioral
intention is provided in Section 2.10 Habit and Behavior. For perceived fit, empirical
research shows a positive relationship between fit and intention (Cable & DeRue 2002;
KristofBrown,Zimmerman&Johnson2005).Therefore,itcanbeanticipatedthatthebetter
thefitbetweenhabitsandtechnology,thehighertheintentiontouseit.Thus,inthisthesis
Hypothesis1expectsthat:

H1:Habittechnologyfithasapositiveimpactuponbehavioralintention.

This thesis also hypothesized that habittechnology fit and behavioral intention would be
moderatedbyage,experienceandgender:

H1a: The relationship of habittechnology fit and behavioral intention will be

moderatedbyage,experienceandgender,suchthattheeffectwillbestronger
forolderandmoreexperiencedmen.

In UTAUT2 (Venkatesh, Thong & Xu 2012) extends UTAUT, among other variables, by
including single predetermined habit. Age, experience and gender were hypothesized as
moderators in the relationship between habit and intention. Age and experience were
empirically confirmed in the same study (Venkatesh, Thong & Xu 2012) where mobile
Internet acceptance was the target technology. Also, a study on electronic banking
70

acceptance (Dabholkar & Bagozzi 2002) has empirically confirmed this moderation effect
withsinglepredeterminedhabit.Thetheoreticalbasesrelatepositthatageandexperience
are connected by time which plays an important role in developing and strengthening
habits.

As time passes peoplegrow olderand gain greater experience with the technologies they
may use. What they have learnt, what it has worked in the past has contributed to build
habitswhichgeneratesresistancetowardschange(Aarts&Dijksterhuis2000;Bargh1990;
Hodgson2010;Limayem,Hirt&Cheung2007;Ouellette&Wood1998;Sheeranetal.2005;
Verplanken&Aarts1999;Wood,Tam&Witt2005).ThisconservativepullofhabitsinWood
& Neal (2009) make older habits more likely to be maintained because of the speed and
easewithwhichpastpatternsofbehaviorcanbeinitiatedandexecuted(Ouellette&Wood
1998).ThiseffectwasfirstdescribedbyDeweyin1922(Dewey2002)whenhereferstothe
weightoftheadultcustomandthehabitsofthegrowingpersontametheoriginalityofthe
youngerandarejealouslykept.

In the case of gender, Venkatesh et al. (2012) suggest that generally women will show
higherlevelsofattentiontodetailcomparedtomen.Basedinstrongarguments(Farina&
Miller1982;Gilligan1982;Krugman1966;MeyersLevy&Tybout1989)aboutdifferencesin
genderandattention,wheregenerallywomenwillshowhigherlevelsofattentiontodetail
comparedtomen.Thisthesisalsofoundempiricalsupportforthatpremisein(Goldner&
Levi 2014), (Iijima et al. 2001), and (Milne & Greenway 1999). Venkatesh et al. (2012)
suggeststhatthegreatertheattentiontodetail,thesmallertheattachmentforonesown
habits. Therefore, if women pay more attention to detail than men, they would be less
attached to their habits. Gender as a moderator of the relationship between habit and
intention has been empirically weakened in a study of age, gender and experience in the
acceptanceof3Dgestures(Comtet2013).Also,thevalidityofthisargumenthasalsobeen
weakenedbytheempiricalVenkateshsresultsinthestudyofmobileInternet(Venkatesh,
Thong & Xu 2012) and a qualitative study of smart mobile technology in financial
informationcommunication(AlHtaybat&vonAlbertiAlhtaybat2013).Still,thishypothesis

71

has rarely, if ever, been tested with nonpredetermined habits. Therefore it was deemed
appropriatetoincludethepropositionofgenderasamoderatorinthisthesis.

Having set the first research question which informed Hypotheses 1 and 1a, this chapter
continuesincursionintechnologyacceptance.Thefollowingsectionshowshowmanyyears
ofresearchwereunifiedinasinglesynthetictheory.

2.22 Theoreticalcontextforhabittechnologyfit:rationaleforitsselection

In previous research of technology acceptance, several competing models have emerged.

Thesemodelsputforwarddeterminantsofacceptancewhichfrequentlyoverlapwithother
determinants suggested by competing models (Cornacchia, Baroncini & Livi 2008). The
unificationofthemostsalienttechnologyacceptancemodels(seeVenkateshetal.2003),
providedasynthesisoftheoverlappingvariableswithanunprecedentedpredictivepower
on behavioral intention. The result was the Unified Theory of Acceptance and Use of
Technology, also known by its acronym UTAUT. The relationships of this theory were
empiricallyconfirmedinextensiveresearch,forinstanceinBandyopadhyayandFraccastoro
(2007) and Venkatesh and Zhang (2010), (see also Table 2.7 Empirical Tests of UTAUT).
UTAUTwasselectedtoprovidetheoreticalcontexttohabittechnologyfitinthisthesis,and
thefollowingparagraphsfurtherjustifythisselection.Acknowledgeofthedownsidesofthis
selectionarealsopresented.

UTAUT was selected because one of most regarded aspects of UTAUT is the synthesis
previous research on acceptance (Chen 2011; Gupta, Dasgupta & Gupta 2008; Reunis,
Santema & Harink 2006; Yeow & Loo 2009); also, because reliability is an important
attribute of UTAUT (Bandyopadhyay & Fraccastoro 2007; Venkatesh & Zhang 2010). Its
predictors are stable and relatively more important than that suggested by the parent
models (Ben Boubaker & Barki 2006). Besides these strengths, UTAUTs author has
highlighted generalizability, predictive validity, and the fact that this theory is technology
adoptionspecific(Venkatesh,Davis&Morris2007a).

72

ThisselectionwasdeemedappropriatedespitetheweaknessesofUTAUT.Incontrasttoits
strengths,UTAUThasreceivedcriticismonimportantaspects.Oneof themostsignificant
commentsonthistheoryisthatithasremovedtherelationshipofattitudesandintention.
Attitudes (Ajzen & Fishbein 1980; Fishbein & Ajzen 1975) are a relevant part of theory to
explainbehavior,andithasbeenconsideredinappropriatetohavethisconstructremoved
based on statistical criteria alone (Ben Boubaker & Barki 2006; Yang 2010; Zhang & Sun
2009). Another shortcoming of UTAUT is the decreasing importance of perceived ease of
use and social influence over time (Karahanna, Straub & Chervany 1999; Taylor & Todd
1995;Venkatesh,Davis&Morris2007a).OthercriticismsalsosuggestthatUTAUT:doesnot
consider cultural dimensions (Srite & Karahanna 2006; Venkatesh & Zhang 2010), ignores
nonutilitarian and hedonistic factors (Bergvik, Svendsen & Evjemo 2006; Yang 2010), and
providesscalesthatmightnotberobustenoughacrosssettings(Gupta,Dasgupta&Gupta
2008;Li&Kishore2006).

These strengths and weaknesses are detailed in the following section. Each of its
componentsisdescribedandexamined.

2.23 TheUnifiedTheoryofAcceptanceandUseofTechnology

The Unified Theory of Acceptance and Use of Technology (UTAUT) is product of a study
conducted by Venkatesh et al. (2003). This study analyzed, tested, compared, and
synthetized 32 beliefs of eight prominent models (Chen 2011): the Theory of Reasoned
Action(Fishbein&Ajzen1975),theTechnologyAcceptanceModels(Davis1989;Venkatesh
& Davis 2000), the Motivational Model (Davis, Bagozzi, & Warshaw, 1992), the Theory of
PlannedBehavior(Ajzen1991),thecombinedTechnologyAcceptanceModelandTheoryof
PlannedBehavior(Ajzen1991;Taylor&Todd1995),theModelofPCUtilization(Thompson,
Higgins & Howell 1991; Triandis 1977), the Diffusion of Innovation Theory (Moore &
Benbasat 1991; Rogers 2003, originally 1962), and the Social Cognitive Theory (Bandura
1986;Compeau&Higgins1995).Afterthesynthesisanewunifiedmodelwasoutlinedas
showninFigure2.3UTAUTModel(Venkateshetal.2003).

73

Figure 2.4 show two structured equations and their graphic representation. The figure
showsbehavioralintentiondeterminedbyeffortexpectancy,performanceexpectancy,and
socialinfluence;whereasactualbehaviorisshowndeterminedbybehavioralintentionand
facilitatingconditions.Behavioralintentionreferstothestrengthofapersonsintentionto
accomplishaparticularbehavior,anditwasdefinedasanindividual'spositiveornegative
feelings (evaluative affect) about performing the target behavior (Davis, Bagozzi &
Warshaw1989,p.984;Fishbein&Ajzen1975,p.288).Intentionhasbeenmeasuredasa
latentvariablewithverysimilaritemsIpredict,Iintend,andIplantouse[technology]in
thenext[periodoftime]asusedin(Davis,Bagozzi&Warshaw1989;Fishbein&Ajzen1975;
Venkatesh et al. 2003). Actual behavior has been defined as a measure of the target
behavior (Davis 1986). Sometimes it has been measured as an observed variable, for
exampleusedurationinsystemlogs(Venkateshetal.2003).But,ithasalsobeenmeasured
asanunobservedlatentvariable;anexampleofthisisthefrequencymeasureusedin(Davis
1986)orasacombinationoftimespentfrequencyofuse,andintensityofuse(Liangetal.
2010).

UTAUTMODEL

74

Figure2.4UTAUTModel (Venkateshetal.2003)

Eachlatentvariablethatdeterminesbehavioralintentionandactualbehaviorisexplained
next. Firstly, a definition of the construct is provided. Secondly, the inheritance from the
parentmodelsisdetailedintheformofnumberofshareditems.Thirdly,fromasampleof
studiestheproportionofsupportingstudiesisreported.

2.23.1 Performanceexpectancy

Performance expectancy (PE) is the degree to which an individual believes that using the
technologywillhelphimorhertoattaingainsinjobperformance(Venkateshetal.2003).
Its measurement items derive from the Diffusion of Innovation Theory (Rogers 2003,
originally 1962) (2 items), Model of PC Utilization (Thompson, Higgins & Howell 1991;
Triandis 1977) (1 modified item), Social Cognitive Theory (Compeau & Higgins 1995) (1).
However, the four items of performance expectancy are contained in the six items of the
Technology Acceptance Model (Davis 1989). This would make UTAUTs performance
expectancyandTAMsperceivedusefulnessfairlyequivalent.Table2.7presentsasampleof
studiesconductedwithUTAUT,inwhichonlyonestudyonknowledgemanagementsystem
(Isabelle&Sandrine2009)reportednotsupportingperformanceexpectancyasdeterminant

75

ofbehavioralintention.Performanceexpectancywassupportedin28otherstudies,i.e.in
96.5%ofthecases.

UTAUT has hypothesized that performance expectancy and behavioral intention is

moderated by age and gender such that the effect would be stronger for younger men
(Venkateshetal.2003).Table2.7showtwentyninestudiesofUTAUT.Thirteentestedage
asmoderatorofperformanceexpectancy,butonlysixsupportedthishypothesis(46.15%).
For gender, fourteen studies tested the hypothesis, only seven the studies supported the
relationship(50%).

2.23.2 Effortexpectancy

Effort expectancy (EE) is defined as the degree of ease associated with the use of the
technology(Venkateshetal.2003).AllfourmeasurementitemsderivefromtheTechnology
Acceptance Model (Davis 1989) (4 items). These items are also share with a quantitative
version (Moore & Benbasat 1991) of the qualitative theory of Diffusion of Innovation
(Rogers 2003, originally 1962) (2 items). UTAUTs effort expectancy and TAMs perceived
usefulnessarethenquiteequivalenttoo.AlthoughconceptuallysimilarwithPCUtilizations
complexity(Thompson,Higgins&Howell1991),noneofitsitemsarerepresentedinUTAUT.
Table2.7showsthatalltwentyninestudiestestedandreportedeffortexpectancy.Twenty
three of the studies supported effort expectancy as determinant of behavioral intention
(79.31%),twomore(6.9%)supportedthehypothesispartially(inseriesofstudies),andfour
didnotsupportedit(13.79%).Thismaysupportempiricalresearchthatpositsadecreasing
importanceofperceivedeaseofuseintime(Karahanna,Straub&Chervany1999;Taylor&
Todd1995;Venkatesh,Davis&Morris2007a).

UTAUT hypothesized that the relationship between effort expectancy and behavioral
intentionwouldbemoderatedbyage,experienceandgender,suchthattheeffectwouldbe
strongerinmoreexperiencedyoungerwomen.Table2.7showsthatage,asamoderatorof
effort expectancy, was tested in thirteen cases. Only seven of those cases supported this
hypothesis (53%). Experience was tested only in seven cases, and five of them supported

76

the hypothesis (71%). In the case of gender, fourteen studies tested it as moderator of
effortexpectancy.Butonlyfivestudiessupportedthehypothesis(35%).

2.23.3 SocialInfluence

Socialinfluence(SI)isdefinedasthedegreetowhichanindividualperceivesthatimportant
othersbelieveheorsheshouldusethenewtechnology(Venkateshetal.2003).Thefour
measurement items of this construct derive from of the Theory of Planed Behaviors the
twoitemsubjectivenorm(Fishbein&Ajzen1975)(2items),andfromPCUtilizationsfour
item social factors (Thompson, Higgins & Howell 1991) (2 items). Image construct (social
image or status) from the Diffusion of Innovation (Moore & Benbasat 1991; Rogers 2003,
originally1962)wasnotrepresentedinUTAUTsmeasurementmodel.Table2.7showsthat
all twenty nine research reports included social influence. Twenty four of those cases
supportedsocialinfluenceasadeterminantofbehavioralintention(83%).

UTAUT hypothesized that social influence and behavioral intention will be moderated by
age,experience,genderandvoluntariness.Theexpectedeffectwouldbestrongerforless
experienced older women in mandatory settings (Venkatesh et al. 2003). Table 2.7 shows
thatonlyelevenoftwentyninestudiesreporttestingageassocialinfluencesmoderator.
Onlythree(27%)supportedthishypothesisforage.Sevenstudiesreporttestingexperience
as a moderator of social influence, six of them supporting the hypothesis (85%). Thirteen
reports present gender as a moderator of social influence, with only five supporting the
hypothesis(38%).Otherthreetestedvoluntariness,andtwoofthose(66%)supportedthe
hypothesis.

2.23.4 Facilitatingconditions

Facilitatingconditions(FC)hasbeendefinedasthedegreetowhichanindividualbelieves
thatanorganizationalandtechnicalinfrastructureexiststosupportuseofthetechnology
(Venkateshetal.2003).ThefinalscaleofUTAUTcontainsfouritemswhichcomefromthe
PC Utilizations (Thompson, Higgins & Howell 1991) (3 items), the Theory of Planned
Behavior (Ajzen 1991) (1 item); whereas the Diffusion of Innovation theory (Moore &

77

Benbasat 1991; Rogers 2003, originally 1962) was not represented in the measurement
model. This shows that UTAUTs facilitating conditions and TPBs perceived behavioral
controlareverysimilarscales.Table2.7showssixteenoftwentyninecasesofstudiesthat
reportonfacilitatingconditionsasdeterminantofactualbehavior.Thirteenofthosecases
(81%)supportedthehypothesis.

UTAUT hypothesized that the relationship between facilitating conditions and actual
behaviorwouldbemoderatedbyageandexperience,suchthattheeffectwillbestronger
forolderindividualsmorewhoexperiencedusersofthetechnology(Venkateshetal.2003).
Table 2.7 presents seven studies that tested age as a moderator of facilitating conditions.
Onlytwo(28%)supportedthehypothesis.Inregardstoexperience,fivestudiesreportedon
thismoderationeffect.Threeofthem(60%)supportedthehypothesis.

2.23.5 InternalhypothesesoftheUnifiedTheoryofAcceptanceandUseofTechnology

Table2.6summarizesUTAUTshypotheses(Venkateshetal.2003),whicharereferredtoas
UTAUTsinternalhypothesestodifferentiatethemfromthehypothesesofthisthesis.These
internalhypothesesarenotgivenanidentifier,suchasH1,H2...,whereasthehypothesesof
this thesis are always identified in this way. The first column presents the independent
variable(X).Thedependentvariablesarebehavioralintention(BI)andactualbehavior(AB).
The first twelve hypotheses correspond to the first, and the last four to the second. For
example,performanceexpectancy(PE)isexpectedtomeetthecriteriaofbeingsignificantly
andpositivelycorrelatedwithbehavioralintention(BI),strongereffectisnotapplicable.In
thecaseofgenderasamoderator(GENxPE),itisexpectedtofindastrongerrelationship
between performance expectancy and behavioral intention in man than in women. The
moderatorisexpectedtobestatisticallysignificantanditscorrelationcoefficientnegative
(alwaysthatmenarecodedas1andwomenas2).

78

INTERNALHYPOTHESESOFUTAUT
ExpectedCriteria
Hypothesis
Effect Linear
XBI Significant StrongerEffect
Direction
1 PE Direct Yes (+) N/A
2 EE Direct Yes (+) N/A
3 SI Direct Yes (+) N/A
4 GENxPE Moderator Yes () Men
5 GENxEE Moderator Yes (+) Women
6 GENxSI Moderator Yes (+) Women
7 AGExPE Moderator Yes () Younger
8 AGExEE Moderator Yes () Younger
9 AGExSI Moderator Yes (+) Older
10 EXPxEE Moderator Yes () Lessexperience
11 EXPxSI Moderator Yes () Lessexperience
12 VOLxSI Moderator Yes (+) Mandatory
ExpectedCriteria
Hypothesis
Effect Linear
XAB Significant StrongerEffect
Direction
13 BI Direct Yes (+) N/A
14 FC Direct Yes (+) N/A
15 AGExFC Moderator Yes (+) Older
16 EXPxFC Moderator Yes (+) Moreexperience
Table2.6InternalHypothesesofUTAUT (Venkateshetal.2003)

79

Table 2.7 presents twentynine studies where UTAUT was tested empirically. As detailed
before, the core determinants of behavioral intention show a tendency to be supported,
whilethemoderatorsshowlesssupport.Anotheraspectshowninthistableisthatmostof
the studies have modified UTAUT by extending it or testing it partially. This is consistent
with Venkatesh and Zhang (2010) who have explained that UTAUT has rarely had been
replicatedfaithfully.Onlytwostudieswerefoundshowingtheanalysisoftheoriginalmodel
beforeextendingit:(Chu2013)and(Venkatesh&Zhang2010).Eleven(38%)outoftwenty
ninestudiesusedregressiontotestthemodel(fourofthoseusedANOVAandregression
13.7%), ten more used PLSthe original technique used in (Venkatesh et al. 2003), and
eightmoreusedSEM(34%).However,theauthorofthisthesiscouldnotfindasinglestudy
testinganunmodifiedspecificationofUTAUTwithSEM.

The study that proposed UTAUT (Venkatesh et al. 2003) used variancebased structural
equationmodelling(PartialLeastSquares)asitsanalysistechnique.PartialLeastSquaresis
anappropriatetechniqueforpredictionandexploratoryobjectives.Thus,thedetermination
coefficient (R2) is usually the goal (Hair, Ringle & Sarstedt 2011). Aligned to its analysis
technique,UTAUTsynthesiswasdrivenbytheaimofincreasingR2.Thefinaloutcomesof
thestudy(Venkateshetal.2003)achievedtoexplain70%ofthevarianceofuseintention
(R2=.70). These levels of R2 have been occasionally attained by slightly modified models
(Bandyopadhyay & Fraccastoro 2007; Venkatesh & Zhang 2010). However, Partial Least
Squaresisnotanappropriatetechniqueforconfirmatoryanalysis.CovariancebasedSEMis
moreappropriate(seeByrne2010;Hairetal.2010;Hair,Ringle&Sarstedt2011).

80

 EMPIRICALTESTSOFUTAUTINPREVIOUSRESEARCH

Specification
PE>BI

FC>AB
GENxPE
GENxEE

AGExPE
AGExEE

AGExFC

EXPxFC
EE>BI

GENxSI

EXPxEE
AGExSI

VOLxSI
SI>BI

EXPxSI
No.

Reference Country Technology Analysis

1 (AbuShanab&Pearson2007) Jordan Internetbanking M S S S S S S NS S NS NS S Regression,ANOVA

1 1
2 (AlGahtani,Hubona&Wang2007) SaudiArabia Useofcomputers M S S S S NS NS NS S NS NS2 NS2 S S NS PLS

Prepaymentmetering
3 (Bandyopadhyay&Fraccastoro2007) India M S S S S S S S S S S S S S S SEM
systems
Internetinnovation
4 (Chu2013) China O S S S S NS NS NS NS NS NS NS S S N N Regression/ANOVA
IntermediaryPlatforms
PictureArchivingand
5 (Duycketal.2010) Belguim M S PS PS S Regresion
CommunicationSystem

6 (Eckhardt,Laumer&Weitzel2009) Germany CVDatabases M S PS PS PS PLS

SocialMediaforHealth
7 (Hansonetal.2011) US M S S S S S Regression
Promotion
Intelligentvital
8 (Hsu,Tseng&Chuang2011) Taiwan M S NS S S Regression
monitoringproducts
9 (Hung,Wang&Chou) Taiwan EGovernmentservices M S S S S Regression
KnowledgeManagemet
10 (Isabelle&Sandrine2009) France M NS NS PS PS SEM
System

(Kijsanayotin,Pannarunothai&Speedie Healthinformation
11 Thailand M S S S S PLS
2009) technology

ActivityBased
12 (Leeetal.2010) Taiwan Costing/Management M S NS S S SEM
System
13 (Lu,Yu&Liu2009) China MobileDataService M S S PS NS NS NS NS NS NS NS NS SEM
Internetbanking
14 (Martins,Oliveira&Popovi2014) Portugal M S S NS NS NS NS NS NS NS NS NS PLS
adoption

15 (McLeod,Pippin&Catania2009) US TaxSoftware M S S S PLS

Largescaleenterprise
16 (Neufeld,Dong&Higgins2007) Canada M S S S S PLS
levelsystem
17 (Ney2013) Netherlands mCRM M S S S NS NS NS NS NS NS Regression/ANOVA
18 (Nistoretal.2013) US VirtualCommunities M S S S Regression
19 (Powelletal.2012) US OnlineVotingSystem M S S S NS NS NS S S S PLS
20 (Schaper&Pervan2007) Austalia HealthICT M S S S PLS

21 (Schaupp,Carter&Hobbs2010) US Efileadoption M S NS S S SEM

22 (AbuShanab,Pearson&Setterstrom2010) Jordan InternetBanking M S S S S S S S S NS Regression,ANOVA

23 (SokFoon&ChanYinFah2011) Malaysia InternetBanking M S S S Regression

Systembeingintroduced
24 (Venkatesh&Zhang2010) US O S S S S S S S S S S S S S S S PLS
intheorganization

Systembeingintroduced
25 (Venkatesh&Zhang2010) China M S S S S NS NS NS NS NS S S S PLS
intheorganization

26 (Wang&Shih2009) Taiwan InformationKiosks M S S S S S NS S NS S NS NS SEM

Distancelearning
27 (Wangetal.2010) Taiwan M S S S S S NS NS SEM
technologies
28 (Wang&Wang2010) Taiwan MobileInternet M S S S S S NS SEM
Multipurposesmart
29 (Yeow&Loo2009) Malaysia M S S S Regression
identityCard
M=Modifiedspecification,O=Originalspecificaiton,S=Supported,NS=NotSupported.
1Findingscontainamisinterpretationinthepresenceofmoderators.Theresultswerereinterpreted.
2Theresultswerereinterpreted,sincethestudyhypothesizedoppositelineardirectiontoUTAUT.

Table2.7EmpiricalTestsofUTAUTinPreviousResearch (Source:Author)
(AbuShanab&Pearson2007;AbuShanab,Pearson&Setterstrom2010;AlGahtani,Hubona&Wang2007;Bandyopadhyay&Fraccastoro2007;Chu
2013;Duycketal.2010;Eckhardt,Laumer&Weitzel2009;Hansonetal.2011;Hsu,Tseng&Chuang2011;Hung,Wang&Chou;Isabelle&Sandrine
2009;Kijsanayotin,Pannarunothai&Speedie2009;Leeetal.2010;Lu,Yu&Liu2009;Martins,Oliveira&Popovi2014;McLeod,Pippin&Catania
2009;Neufeld,Dong&Higgins2007;Ney2013;Nistoretal.2013;Powelletal.2012;Schaper&Pervan2007;Schaupp,Carter&Hobbs2010;Sok
Foon&ChanYinFah2011;Venkatesh&Zhang2010;Wangetal.2010;Wang&Wang2010;Wang&Shih2009;Yeow&Loo2009)

81

2.24 Researchquestion2

Since habittechnology fit is a new construct which addresses the gap of multiple non
predetermined habits in the literature of habit (see Section 2.2 and Table 2.1), other
relationshipsofhabittechnologyfitareunknowninconsequence.Singlehabitshavebeen
studied in the theoretical framework of UTAUT. However, multiple nonpredetermined
habitsmayhaveadifferenteffectthanhabits.Particularly,habittechnologyfitmayhavea
different effect than single habit when it is included in a model. Another significant gap
identifiesthatUTAUThasrarelybeentestedsimultaneouslyusingconfirmatorytechniques
of analysis (such as covariancebased SEM) in unmodified specifications (see Table 2.7
Empirical Tests of UTAUT in Previous Research). Covariancebased SEM would provide a
valueofstatisticalsignificancetosupportorrejectatheoreticalmodelasawholewhichto
dateisyetunknown.Asaconsequence,theeffectofhabittechnologyfitinthestructural
model of UTAUT is to be identified. The second research question addresses the gap of
multiple nonpredetermined habits and implicitly the gap of confirmatory analysis for
UTAUTasfollows:

RQ2WhatistheeffectofincludingHabitTechnologyFitintheUnifiedTheory
ofAcceptanceandUseofTechnologymodel?(H2&H3vsH4&H5)

Inordertogiveananswertothisquestion,itisnecessarytocompareUTAUTvsUTAUTplus
habittechnologyfit.ThisnecessityiswhatitinformsHypotheses2to5.

2.25 Hypotheses2and3

Hypotheses2and3arebasedontheassumptionthatUTAUTisavalidtheoreticalmodelin
two ways (this has been empirically supported to some extent by previous research, see
Sections 2.23.1 to 2.23.4 ). Firstly, its variables should achieve criterionrelated validity,
which is the basis of construct validity (DeVellis 2012). Secondly, the model as a whole
wouldbeexpectedtofitwiththedataandachievestatisticalsignificanceforsuchfit(Byrne
2010; Hair et al. 2010; Hair, Ringle & Sarstedt 2011). Hypotheses 2 and 3 reflect these
assumptionsrespectively.

82

InregardstothefirstwayinwhichUTAUTisexpectedtobeavalidmodel,criterionrelated
validityisachievedonceascalehassuccessfullyprovedassociationwithsomecriterionor
standard.Criterionrelatedvalidityisoftenreferredaspredictivevalidity,andthetraditional
indexofcriterionrelatedvalidityisthecorrelationcoefficient.Theprobabilisticvaluecanbe
used as an accuracy criterion. Criterionrelated validity is neutral and does not imply
causality; however it can directly strengthen or weaken construct validity which is
concerned with causality. Construct validity cannot be established if criterionrelated
validityisnotestablishfirst.Theybothcanbeassessedbytheextenttowhichameasure
behavesthewayitisexpected(Cronbach&Meehl1955;DeVellis2012;Ghiselli,Campbell&
Zedeck1981).

Asexpectedinstrongtheory,Hypothesis2anticipatestherelationshipsbetweenvariables
hypothesizedinUTAUT(Venkateshetal.2003)willbeobserved:

H2: The original models internal hypotheses will achieve criterionrelated

validity.

In relation to the second way in which UTAUT is expected to demonstrate validity,

covariancebased SEM tests the complete specification of a model simultaneously.
Therefore,itisusefultoconfirmcompletetheoreticalrelationships.SEMevaluatestowhat
extent data fits with the theoretical model and how likely it is to find good fit in other
samplesofthesamepopulation(Byrne2010;Hairetal.2010;Hair,Ringle&Sarstedt2011).
Evaluating the unmodified UTAUT model simultaneously by confirmatory analysis
techniques (covariancebased SEM) may addresses a gap in the literature of technology
acceptance.Asitisnecessarytolaterestimatetheeffectofintroducinghabittechnologyfit
inUTAUT,theevaluationofthisHypothesisconstitutesapremisewhichdirectlycontributes
toanswerResearchQuestion2:

H3: The original model will have an acceptable fit with the data and will be
statisticallysignificant.

83

2.26 Hypotheses4and5

ResearchQuestion2requirestheevaluationofunmodifiedUTAUT,aswellastheextended
UTAUTinordertocomparethemanddeterminetheeffectofintroducinganewvariable.In
thatwayResearchQuestion2informsHypotheses4and5,astheyservetodeterminethe
effectofextendingUTAUTbyincludinghabittechnologyfit:

H4: The extended models internal hypotheses will achieve criterionrelated

validity.

H5: The extended model will have an acceptable fit with the data and will be
statisticallysignificant.

Once more, Hypothesis 4 suggests criterionrelated validity of the extended model, which
canbetestedandcomparedtothetestofHypothesis2.Similarly,Hypothesis5positsthe
extendedmodelisatheoreticallyvalidmodel,anditstestcanbecomparedtothetestsof
Hypothesis3.Oncecompared,theeffectofincludinghabittechnologyfitcanbeevaluated
at the level of individual relationships or at the level of the complete model. Then, an
answertoResearchQuestion2canbeprovided.

2.27 ResearchQuestion3

Singlehabitasaconstructisproposedinsufficienttouncovertheroleofmultipleothernon
predeterminedhabits,whichisextensivelyunknown;thereforeitnecessaryanewconstruct
such as habittechnology fit that attempts to capture the explanatory power of multiple
habits (see Table 2.1 Gap in Literature). The effect of habittechnology fit in technology
acceptanceandparticularlyinUTAUTisstillundiscovered.Itisalsouncertainiftheinclusion
ofthenewconstructcouldimproveordeterioratetheUTAUTsmodelfitortheexplanatory
power of its determinants upon behavioral intention or actual behavior. A new construct
opensnewpossibilitiestomodellingtheory,andthisiswhyathirdresearchquestionwas
formulated:

RQ3Ifitcould,howcanhabittechnologyfitimprovethemodelspecification
of the Unified Theory of Acceptance and Use of Technology? (Posthoc
modification)

84

ResearchQuestion3cannotbeanticipated.Therefore,nohypothesescanbederivedfrom
thisquestion.However,itcanbeapproachedwithposthocmodelmodification.

2.28 Conceptualframeworksummarized

Figure 2.5 summarizes and integrates some of the important concepts that have been
review in this chapter. On the top of the figure the circles represent roughly the diverse
modelsintechnologyacceptanceresearch.

CONCEPTUALFRAMEWORK

Figure2.5ConceptualFramework (Source:Author)

Therearesomewhichhavebeenseminalinthehistoryofthefield;thereareothersthat
simplyremainunnamed(m1,m2mn).AmongtheseminalworkistheTheoryofReasoned
Action, Theory of Planned Behavior, Technology Acceptance model, etc. The models
included in Figure 2.5 are not meant to be precise about the overlapping portions or the
explanatorypowerofeachmodel.However,itrepresentstheideaofseveralcompetingand
overlappingmodels(Cornacchia,Baroncini&Livi2008).Allthesemodelshavebeentrying
85

to explain fairly the same underlying concept (Venkatesh et al. 2003). UTAUT has just
synthesizedthemostsalientmodelstosomeextent(Chen2011;Gupta,Dasgupta&Gupta
2008;Reunis,Santema&Harink2006;Yeow&Loo2009).

On the left side of Figure 2.5, Ht represents the habit that corresponds to use the target
technology.Hthasbeenidentifiedandithasbeenincludedinsomeoftheseminalmodels,
see(Honkanen,Olsen&Verplanken2005)and(Saba,Vassallo&Turrini2000)extendedTRA
andTPBwithhabit,(Wu&Kuo2008)and(Gefen2003)diditwithTAM,(Huang,Wu&Chou
2013) with the TaskTechnology Fit model, and (Pahnila, Siponen & Zheng 2011) and
(Venkatesh, Thong & Xu 2012) with UTAUT. The circles labeled H1, H2 Hn, represent the
structureofindividualhabitswhicharedevelopedalonglife.Thesehabitsdonotexistisa
pure isolated form. They are rather a structure which defines taste, choices, and
understanding(Bourdieu1984;Swartz2002;Wozniak2009)thereforethereactiontowards
technology, the intention to use it, and the actual behavior of utilization (shown at the
bottom of the conceptual model). The dotted lines from the habits to the three boxes
describe some of the causal habits of which the individual might not be aware. But the
green dotted box suggests that some unaware habits can be brought to consciousness
retrospectivelybytracesofaction(Mittal1988).Thehabitscontainedinthegreenboxare
thosethatcouldpossiblybejointlycapturedwithperceivedfit.Asbothhabitandperceived
fit are considered significant determinants of behavioral intention (Cable & DeRue 2002;
KristofBrown,Zimmerman&Johnson2005),itcouldbeanticipatedthatthebetterthefit
betweenhabitsandthetechnology,thehighertheintentiontouseit.

2.29 Researchmodel

Figure 2.6 shows the model that will be tested in this research. The circles indicate the
unobserved or latent variables and the boxes represent observable measures. The arrows
representacausalrelationshipbetweenvariablesthathasbeenhypothesized.Thosearrows
running from one circle to another represent direct relationships, whereas the arrows
runningfromacircletosomepointinthemiddleofanotherarrowrepresentmoderating
effects. There are blackthin line arrows and greenthick arrows. The first kind represents

86

UTAUTs internal hypotheses. The second type represents the original contribution of this
thesis,andrepresentsHypotheses1and1a:

RESEARCHMODEL

Figure2.6ResearchModel (Source:Author)

H1:HabitTechnologyFithasapositiveimpactuponbehavioralintention.

H1a: The relationship of habittechnology fit and behavioral intention will be

moderatedbyage,experienceandgender,suchthattheeffectwillbestronger
forolderandmoreexperiencedmen.

All UTAUTs internal hypotheses represented by thin black line circles and arrows are
consideredtogetherinthisthesis,andrepresentedinHypotheses2and3:

H2: The original models internal hypotheses will achieve criterionrelated

validity.

87

 H3: The original model will have an acceptable fit with the data and will be
statisticallysignificant.

Thesumoftheblackthinandthegreenthickfigures,i.e.thefullmodel,arerepresentedin
Hypotheses4and5:

H4: The extended models internal hypotheses will achieve criterionrelated

validity.

H5: The extended model will have an acceptable fit with the data and will be
statisticallysignificant.

Thelabelsinthecirclesstandfor:habittechnologyfit(HTF),performanceexpectancy(PE),
effortexpectancy(EE),socialinfluence(SI),facilitatingconditions(FC),behavioralintention
(BI)andactualbehavior(AB).

2.30 Summary

This chapter reviewed literature from three fields. First, it reviewed theory about habits.
Secondly, it explored the concept of fit in personenvironment fit literature. Thirdly, this
chapterrevisited literature on technology acceptance with focus on the Unified Theory of
AcceptanceandUseofTechnology.Thischapteridentifiedimportantgapssuchasthestudy
of multiple nonpredetermined habits in technology acceptance. A new construct was
conceptualizedtoaddressthisgap.Habittechnologyfitwastheorizedcombiningperceived
fitandhabits.Threeresearchquestionsandsixhypotheses(H1,1a,25)weredeveloped.
Thischapteralsopresentedaconceptualframeworkandtheresearchmodelofthisthesis.

88

 CHAPTER3 METHOD
3.1 Objective

Thepurposeofthischapteristojustifyandexplainthemethodologyandresearchdesign
usedtoaddresstheresearchquestions.Thischapterincludestwomethodologicalphases.
Firstly,itexplainsthescaledevelopmentofhabittechnologyfit.Secondly,itdescribesthe
researchdesignandtheprocessofdatacollection.Thefirstpartofthischapterwilldiscuss
the research design, the second part the scale development, and the last the statistical
methodsusedinthisstudysuchasfactoranalysesandstructuralequationmodeling.

3.2 ResearchProcedureOverview

Theresearchprocedureofthisthesiscanbesummarizedinfivestages,showninFigure3.1.
Thesestagesincludeeveryaspectreviewed,analyzed,developedandreportedduringthe
researchprocess.

The first stage began with the literature review of habit, personenvironment fit and
technology acceptance. It included identifying research gap, the main theoretical models;
developingofaconceptualframework,researchquestions,andhypotheses.

The second stage included measurement, questionnaire, sample frame development, and
approval from the Ethics Committee. Then, a pretest study was conducted to refine the
instrument,followedbyapilotstudy.

Thethirdstageconsistedinthedeploymentofmainstudy.Anonlinesurveywasconducted
withrespondentsfrom25countries.

The fourth stage, data analysis, included processes of data preparation, reliability tests,
exploratoryandconfirmatoryfactoranalyses,validityassessment,modeltestingandpost
hocmodelmodification.

The fifth stage returns to the literature review to give interpretation to the findings, and
finallyreporttheactivitiesconductedinthisresearch.

89

 STAGESINTHERESEARCHPROCESS

Figure3.1StagesintheResearchProcess (Source:Author)

3.3 Researchparadigm

Atheoreticalperspectiveprovidesaframeworkandalanguagetoexplainsociallife.There
are three dominant paradigms in social research: positivist, interpretative and critical
(functionalism, interpretivism, Marxism) with several variants. This thesis has deemed
appropriatetofollowapostpositivistapproach,associatedwithanobjectiveapproachto

90

thestudyofsocialreality,whichcanonlybeimperfectlyandprobabilisticallyapprehendable
(Blaikie2010;Guba&Lincoln1994;Neuman2010).

Positivist, interpretative and critical paradigms include ontological and epistemological

assumptions,andassociatedpracticesinpursuingknowledge.Ontologicalassumptionsrefer
towhattheworldlookslikeandhowitworks.Epistemologicalassumptionsrepresentthe
perspectivesonhowknowledgecanbedeveloped.Positivistsassumethatrealityexistsand
canbeapprehendedwithanobjectiveapproach.Thepositivistmethodologyisassociated
withexperiments,hypothesisverificationandquantitativemethods.Postpositivismderives
from positivism. Still, it believes in a real reality, but it can only be captured imperfectly
and probabilistically. Postpositivism broadens its methodology and it can even embrace
qualitativemethods.Ittakesaperspectivearguingthattruthcannotbeverified,insteaditis
possible to reject false hypothesis (Guba & Lincoln 1994). The interpretive approach sees
reality as relative; it seeks to explain the subjective meaning of social action with
interpretativedialecticalmethodology(Bryman&Bell2007).Thecriticalapproachstandson
historicalrealism,whereavirtualrealityisshapedbysocioeconomiccharacteristics(Guba&
Lincoln 1994). In the critical approach, the process of inquiry is participative and aims
changingtheconditionsoftheworld(Neuman2010).

3.4 Researchmethodologyjustification

Thisthesishastakenapostpositivistquantitativeapproachtoinvestigatetherelationship
between habittechnology fit and behavioral intention in the context of technology
acceptance. The reason is that the research questions of this thesis derive from previous
empirical research with quantitative tradition: the Theory of Reasoned Action (Fishbein &
Ajzen 1975), the Technology Acceptance Models (Davis 1989), and the Theory of Planned
Behavior (Ajzen 1991) on the side of technology acceptance. But, it also has opted to
developanewconstructbasedonperceivedfitwhichKristofBrownandBillsberry(2012)
havedirectlyaffirmedbelongstoapostpositivistperspective.Thisthesisconcentratedon
the impact of habits upon behavioral intention, and found a gap in the literature where
wider structures of habits are not being considered in the measurement of habit.

91

Measurement, numerical data and hypothesis testing are the core of answering the
researchquestionsofthisthesis,whichjustifiesthequantitativeapproach.

Postpositivism can even embrace quantitative and qualitative methods. It takes a

perspectivewhicharguesthattruthcannotbeverified,insteaditispossibletorejectfalse
hypothesis (Guba & Lincoln 1994). However, advantages and disadvantages are
acknowledged in both approaches. Quantitative methods allow answering prespecified
research questions that require tightly structured designs and prestructured data.
Quantitativemethodsareidentifiedwithdeductivestrategiesofresearch,whichbeginwith
tentative theory, deduce hypotheses and test them (Blaikie 2010; Crowther & Lancaster
2012). In contrast, qualitative methods work well for general guiding questions, loosely
structured designs and not prestructured data (Punch 2005). Qualitative methods are
identified with inductive strategies of research, which start with data collection, continue
with analysis to produce generalizations. These generalizations are propositions that may
becomelawlikewithfurthertesting(Blaikie2010).

Quantitative method finds its strength in that results can be inferred to the rest of the
population, and the costs of distributing, collecting and capturing questionnaires can be
kept low. However, this can also become very costly and time consuming (Brand 2003).
Quantitativemethodsarelimitedtomeasurethedegreeandextentofattitudes,whereas
qualitative methods provide indepth understanding of attitudes and behaviors
(Ledgerwood & White 2006). Qualitative methods provide a narrative offering better and
richer understanding of behaviors (VanderStoep & Johnston 2009). However, results can
hardlybeinferredanditsinterpretativenaturecanintroducebiaswhilecollectingdataor
duringtheanalysis(Brand2003).

Since the research questions of this thesis indicate the adequacy of pure quantitative
methods,thedisadvantagesofthisapproachhastobeaddressed.Inthecaseofthisthesis,
thebiasproducedbythequestionnairewasminimizedinthefollowingways:thecontent
validityoftheitemswasrevisedbyexpertsandconfirmedinaQsortingexercises(DeVellis
2012;Nahmetal.2002).Thequestionnairewascarefullyshapedbasedonliterature,and
theseexpertconsultationandQsortingexercises.But,itwasalsorevisedbythirdpartiesin
92

apreteststudy.Expertsandtherevisersinthepretestwereexplicitlyaskedtoindicateany
aspectthatcouldbiastheanswers(Stopher&Metcalf1996).Tokeeptheuseofresearch
resources within the budget and timeframes, online surveys were used. The online
surveying system also allowed skipping manual capture of the data, reducing potential
errors(Fink2009).

Other research decisions were made for this research, such as the sources of data (semi
naturalsocialsetting),typedata(primary),samplingstrategy(nonprobabilitysample),type
of design (crosssectional), and data analysis techniques (primarily, structural equation
modelling).ThesedecisionsarediscussedinSection3.5.

3.5 Researchdesign

The activity of research design involves decision making about the different aspects of a
research project (Blaikie 2010). According to Neuman (2010), designing a study includes
decisions about the type of sample, measure of the variables, and data collection
techniques(suchasquestionnaires,experiments).ForBlaikie(2010),thesedecisionsbegin
fromselectingatopicandendwithdecisionsofdataanalysis.Thedesignofthisresearchis
given in four areas: data collection technique, sampling strategy, measurement of the
variables,andanalysistechniques.Then,theresearchprocedureispresented.

3.6 Datacollectiontechnique

Theparadigmofresearchhasbeendefinedaspostpositivistandtheapproachtoanswer
the research questions quantitative. There are four quantitative techniques in social
research. These are structured observation, selfadministered questionnaire, structured
interview and content analysis of documents (Blaikie 2010). Structured observation is
appropriate for artificial settings and experiments and content analysis of documents is
appropriateforresearchquestionswherethemainfocusissecondarydata.Therefore,self
administeredquestionnairesandstructuredinterviewsaretheoptionsofthisresearch.

Selfadministered questionnaires can be posted by regular mail, provided onsite, or

distributed online. Mailed questionnaires advantages are that they can reach large
93

geographicareas,agreaternumberofpeoplecanrespondwithapencil,andthesurveycan
becompletedanywhere.Itsdisadvantagesareitneedsamotivatedsamplewillingtosend
thequestionnaireback,andrespondentsmustbeabletoreadandwrite.Anuptodatelist
ofaddressesisnecessary.Someofthecostsmayincludetheprintouts,envelopes,stamps
and incentives. Onsite handing out offers the benefit of getting immediate response,
respondents can ask questions about the survey, and in some cases surveys can be
conducted in groups. Onsite handing out is limited to the people who are in the place,
respondentsmustreadandwrite.Spacewithsomeprivacyisneededfortherespondents.
Thecostmayincludetheprinting,incentives,asurveysupervisor,andaspace.Onlineself
administered questionnaires are Internet based, therefore their advantages include global
reach,theorderofthequestionsaredeterminedbytheresearcher,onlytheanswersthat
fall within the standards are accepted, help can be provided with hyperlinks, data are
automaticallyenteredandmightbeautomaticallyanalyzed.Someofthedisadvantagesare
that respondents must have Internet access, they may experience technical problems
relatedwithbrowsercompatibilityorserverreliability,andforsomeusersprivacymightbe
aconcern.Thecostsofonlinesurveysaretechnicalsuchastheneedforasurveydesigner
andthesoftwaretocreateanddistributethequestionnaire(Fink2009,p.9).

Structured interviews are usually administered facetoface or over the telephone. The
advantagesarethattheyallowassistancewithunfamiliarwordsorexplorationofanswers
withtherespondent.Thedisadvantagesinbothcasesarethatatrainedintervieweranda
placearenecessarytoconducttheinterview.Adisadvantageofthefacetofaceinterview
might be the risks associated with dangerous interviewees. The telephone interview may
require paying for incentives, telephone charges, computers and technical expertise. The
facetoface interview costs may include training, spaces for the interview, and incentives
(Fink2009,p.9).

Selfadministeredquestionnaireswereselectedforthisresearch.Firstly,thistechniquewas
selectedbecauseitisalignedwiththeresearchquestionandtheresearchparadigmofthis
thesis,andsecondlybecauseresearchintechnologyacceptancehastraditionallyusedself
administeredquestionnaires,see(Ajzen&Fishbein1980),(Davis1989),and(Venkateshet

94

al. 2003). Thirdly, online distribution was selected for this research because this study
addressesInternetusers;bydistributingonlinequestionnairesthetargetsampleisreached.
Besides, it was considered important to reduce errors in entering data, avoid incomplete
questionnaires,andkeeplowcostsforthisresearch.

In order to minimize the disadvantages of an online survey, the researcher took the
following actions. A robust SoftwareasaService surveying system was selected
(qualtrics.com).Thissoftwareishostedinlargedistributedandprofessionallymanageddata
centers.Thatmeansthesystemdowntimeofthesystemisnegligible;datacentershavethe
resources to maintain the highest security standards to safeguard the privacy of
respondents.Mostwebbrowsers(oldandnewversions)areprofessionallytestedtoensure
they are supported in order to avoid compatibility issues. Besides, RMIT University has a
contractwithQualtricsthatcontemplateanypossibleconcernoftheRMITUniversityEthics
Committee.

3.6.1 Datacollectionandtiming

In regards to time, there are two designs to consider: crosssectional and longitudinal. In
crosssectional designs different people are investigated at the same time, whereas in
longitudinaldesignsonegroupofindividualsisexaminedastheydevelop(Kail&Cavanaugh
2011).Crosssectionalstudieshavetheadvantageofbeingtimeefficient,andtheydonot
requirethecooperationofrespondentsoveralongperiod.Itsdisadvantageisthatitcannot
assess changes in individuals over time. Longitudinal designs can assess those changes in
individuals,buttheyaretimeconsumingandaportionoftheparticipantswillusuallydrop
out,whichmaybiasresults(Gravetter&Forzano2010;Kail&Cavanaugh2011).

Crosssectionalstudiescaptureastillpictureforthevariablesincludedintheresearch.This
choice is appropriate for aspects of social life like demographic characteristics, attitudes,
beliefsandbehavior.However,itisnotwellsuitedforprocessesthathappenovertimelike
social change. Longitudinal designs are appropriate when research requires capturing
changesintime,forexampletoevaluatetheimpactofinterventions(Blaikie2010;Donley
2012). Research on technology acceptance has been conducted in mainly crosssectional

95

andlongitudinalstudies,see(Lu,Yao&Yu2005)and(Davis&Venkatesh2004).However,
perceived ease (effort expectancy) of use and social influence in technology acceptance
haveshowndecliningimportanceovertime(Karahanna,Straub&Chervany1999;Taylor&
Todd 1995; Venkatesh, Davis & Morris 2007a), which would make longitudinal studies
desirable.

Acrosssectionaldesignwaschosenbecauseofthetimelimitationoftheresearchproject.
Besides,forthepurposeofthisresearch,changesintimearenotpartofthescopeofthis
thesis.Possiblechangesintimefortherelationshipshypothesizedinthisthesismayoccur,
andtheirassessmentisalimitationofthisresearch.

3.7 Samplingstrategy

The population of this study has been defined as: SoftwareasaService (SaaS) users in
publicclouds.Therationaleisthatasoftwareproducerwhocreatesapieceofsoftwareto
beprovidedasaservicereachesaglobalaudience(PwC2013).SaaSmodelandpubliccloud
deploymentareubiquitousbydefinition(Mell&Grance2011).However,suchaglobaland
ubiquitouspopulationhasimplicationsinthesamplingstrategyofthisresearch.

The most important decision in sampling is between probability and nonprobability

samples. Probability samples have the advantage of being effective in the absence of
detailed information about a universe of elements, they tend to be unbiased, and their
estimatesprecise.Somedisadvantagesarethattheyrequireskillfuldesigners,requirealong
timeforplanningandexecution,andhavelargercostscomparedtononprobabilitysamples
(Sharma2005).Thebenefitsofnonprobabilitysamplesarethattheproceduresareeasier,
quicker and less costly than probability sampling. However, these samples cannot draw
inferences about the population (Blaikie 2010). Although, probability samples are more
appropriate,nonprobabilisticsamplingisselectedincaseswhere:thepopulationcannotbe
identified, it would be very costly, or a list of the elements is not available. Particularly,
purposive and snowball sampling are appropriate for when the population can hardly be
identified(Blaikie2010).

96

The population of SoftwareasaService (SaaS) users in public clouds would be nearly

impossibletoidentify.Thereisnolistofthepeoplewhobelongtothepopulationwhichis
necessary to attempt to conduct probability sampling. Therefore, this research selected a
nonprobabilitysamplingstrategy.

In order to mitigate this limitation, this thesis adopted the RespondentDriven Sampling
method (RDS). RDS consists of a chainreferral sampling approach where recruiters and
recruits influential characteristics are recorded (Heckathorn 2002). Then the statistical
estimation of the sampling bias by homophily is calculated. Heckathorn suggests, despite
bias, indicators computed from chainreferral sampling data can mitigate the probabilistic
disadvantage.Demographicvariablessuchasage(seeNey2013;Powelletal.2012),gender
(AbuShanab,Pearson&Setterstrom2010;Wang&Shih2009)andlevelofindividualismof
culture(Bandyopadhyay&Fraccastoro2007;Martins,Oliveira&Popovi2014)havebeen
identified in the literature of technology acceptance as influential characteristics in
adoption.Thus,thisresearchattemptedtogiverepresentationtoawidespectrumofages,
assure gender balance, and have balance of participation from countries identified as
individualistsandcollectivists.

Inarangeof1to0,homophilyforthedemographicsofthisthesisssamplewaslessthan
0.35 in all cases. The maximum of +1 indicates that all ties between the person who
recruitedandtherecruitswerewithinthesamegroup(perfecthomophily)whilezerowould
indicate no homophily. The selected characteristics for homophily measurement were
gender,age,individualismindexofthecountryoforigin.

The sampling strategy selected may represent a limitation of this research. This is
particularly,becausefromthetargetpopulation,thisthesiswillonlyaddressadultswhocan
understandEnglishtoanswerthequestionnaire,andrepresentativenesscannotbeverified.
However, aligned with the postpositivist perspective, falsifying hypotheses would also
contribute to knowledge (Guba & Lincoln 1994), and generalization of the findings may
requirefurthertestingwithothersamples(Blaikie2010).

97

3.7.1 Unitofanalysis

Units of analysis are wholes that researchers distinguish and can treat as independent
elements,thewhomorwhatunderstudy(Babbie2013;Krippendorff2012). Thisthesis
hasdefineditsunitofanalysisascomprisedofadultswhoareSoftwareasaServiceusers
inpublicclouds,andcanunderstandEnglish.

3.7.2 Samplesizeandresponserate

Foranalysisofthemostcomplexmodelspecificationinthisstudy,theminimumsamplesize
recommendedforthecharacteristicsofthisstudyis =210.Thiswascalculatedbasedon
=50+(8)x(numberofpredictors)(Tabachnick,Fidell&Osterlind2001).Therefore,the251
respondentsrecruitedinthemainstudywasconsideredappropriate.

The questionnaire was distributed to 1,433 people, with 251 adult respondents in 25
countriesprovidedusablequestionnaires.Slightlymorefemales(54.6%)thanmales(45.4%)
participated.Aresponserateof17.51%wascalculated,accordingtotheformulasuggested
by Neuman (1994) and Saunders et al. (2011). More descriptive details of the sample are
presentedinChapter4Table4.1DescriptiveStatisticsoftheSample.

3.8 Measurementofthevariables

Thereareseveralscalestomeasureattitudesandbeliefs.ButaccordingtoPeterson(2000),
three of them are quite influential in measuring and scaling: Likert scales, semantic
differential scales, and Stapel scales. Likert scales (Likert 1932) evaluate statements on a
scaleofagreement;theyareusuallyafivepointratingthatrangesfromstronglyagreeto
stronglydisagree,butalsoseventoelevenpointcanbeused(Nunnally&Bernstein1994).
SemanticdifferentialscalesweredevelopedbyOsgoodetal.(1957),andconsistofaseven
category rate around bipolar adjectives. Stapel scales (Stapel 1969) are a 10category
unipolarrating,rangingfrom5to+5aroundasingleadjectiveinthecenter.

The advantage of Likert scales is they are easy to create, but its disadvantage is that it is
hardtointerpretthemeaningofasinglescore.Thebenefitsofsemanticdifferentialscales

98

are that they are easy to create and allow comparison. Their disadvantages include that
appropriateadjectivesmustbefound,anddatacanbeordinalbutnotinterval.Staplescales
areeasiertocreateandadministerthansemanticdifferential,howeverthedisadvantageis
that they might be more difficult to interpret because their extremes are numerical not
words(Zikmund2003,p.326).

ThisthesisselectedasevenpointLikertscaleinordertomeasurethelatentvariables.The
reasonwastomaintainconsistencywithoriginalUTAUTscales(Venkateshetal.2003).The
disadvantageinregardstointerpretationofasinglescoreinherenttothiskindofscalewas
minimizedbytheimplicitpurposeofthetechnologyacceptancestudies.Theinterestisnot
uponthedescriptiveinterpretationofitsvariables,butonthecovarianceoftheirscales,see
(Fishbein & Ajzen 1975), (Davis 1989), (Venkatesh & Davis 2000), (Taylor & Todd 1995),
(Ajzen 1991), (Triandis 1977), (Thompson, Higgins & Howell 1991), (Moore & Benbasat
1991),(Bandura1986),(Compeau&Higgins1995),and(Venkateshetal.2003).

3.8.1 Measurementprocedure

The study included seven latent variables such as habittechnology fit (HTF), performance
expectancy(PE),effortexpectancy(EE),socialinfluence(SI),facilitatingconditions(FC),and
behavioralintention(BI).Theselatentvariablesweremeasuredusingatleastthreetofour
itemsthatcouldreliablybereflectiveoftheconstruct.SevenpointLikertscaleswereused
inallcases.ThesescalesrangedfromstronglyDisagree(1)tostronglyagree(7).Except
for habittechnology fit, all the measurement items were taken from Venkatesh et al.
(2003). Table 3.3 Measurement Items and Demographic Questions, shows what it was
includedinthesurvey.

Actual behavior, voluntariness (VOL) and experience (EXP) are treated as measures in
UTAUT (Venkatesh et al. 2003). Therefore, the same treatment is applied in this study,
exceptforactualbehaviorwhichwasmeasuredasalatentvariableasin(Liangetal.2010)
because system logs were not available. Voluntariness was evaluated using a seven point
scale (Completely Nonvoluntary (1)/Completely Voluntary (7)), whereas experience was
selectedfromthefollowingoptions:Ihaveneverusedit,6monthsorless,1year,2years,3

99

years,4years,5yearsormore.Thequestionnaireincludeddemographicquestionssuchas
age,gender,countryoforigin,highestlevelofeducation.

Measurement is always a concern in research. Adopting unsystematic measurement

approaches increases the risk of producing inaccurate data. Therefore, the scale
developmentprocessinthisresearchfollowedathoroughprocessbasedontheguidelines
suggestedinDeVellis(2012)book ScaleDevelopment:TheoryandApplicationsandDavis
(1989)scaledevelopmentfortheTechnologyAcceptanceModel(TAM).

3.9 Measurementdevelopmentprocedureforhabittechnologyfit

AccordingtoDeVellis(2012)theprocessconsistsofsixsteps(seeFigure3.2).Thefirststep
isconceptual;theresearcherhastodetermineclearlywhatistobemeasured.Thisstepwas
takeninSection2.18Definitionofhabittechnologyfit.Thesecondstep,theresearcher
has to generate an item pool and determine the format for measurement. The
measurementformatwasdeterminedbeforeinSection3.8matchingUTAUTsformats,and
the procedure for item generation will be presented in the next section (3.9.1 ). In the
fourthstep,theinitialpoolofitemshastobereviewedbyexpertsandtheitemshavetobe
validated.TheexpertconsultationprocedurewillbepresentedinSection3.9.2.Validation
of the items is provided by semantic clustering presented in Section 3.9.3 QSorting
Exercise. The fifth step requires the items be administered to a development sample, so
theresearchercanevaluatetheitems.ThisprocedurewasperformedduringthePilotStudy
presented in Section 3.12 . Finally, in the sixth step the length of the scale has to be
optimized. The optimization was performed after the data collection of the main study in
Section4.4ExploratoryFactorAnalysis.

3.9.1 Itemgeneration

Aconstructanddefinitionforhabittechnologyfitwasexpressedasthedegreetowhichan
individual believes that using the technology is compatible with his or her habits. A scale
was developed for the new construct as well, designed to be used in the context of
technologyadoption,andparticularlytobemeasuredandtestedintheUTAUTmodel.

100

 MEASUREMENTDEVELOPMENTPROCEDURE

Figure3.2Measurementdevelopmentprocedure Basedon(DeVellis2012)

The new scale was designed consistent with other scales of the measurement model of
UTAUTtakenfrom(Venkateshetal.2003).Therefore,theformatofthemeasurementwas
keptasevenpointLikertscale(StronglyDisagree/StronglyAgree).Theresearchergenerated
twenty sentences. The additional items were generated for the original scales, but it was
laterdecidedtokeeponlytheoriginalitems.

DeVellis(2012)explainsthatitemsshouldreflectthescalespurpose.Throughvariationsin
grammatical structure and choice of words, the researcher should attempt to capture the
phenomenon of interest. In the process of item generation each item was individually
assessed for clarity. Three key ideas were systematically checked in each item: each item
had to assess the compatibility with the use of technology, include general habits as
opposed to one specific habit, and the item should be applicable to scenarios where the
respondenthadneverusedthetechnologyorhadextensivelyusedit.Eachitemwasrevised
againstthedefinitionsoftheotherconstructsinUTAUTtoavoidconceptualoverlap.

101

Bytheendoftheitemgenerationandinitialassessment,20itemswerereducedto10.The
eliminationwasbasedonthecriteriapresentedinthetwopreviousparagraphs.Onlyitems
thatcompliedwithalltherequirementscontinuedtothenextphaseofexpertconsultation.

3.9.2 Expertconsultation

The researcher approached four experts from different but relevant fields. They were
selectedbecauseoftheirexpertiseincommunication,linguistics,semioticsandinformation
systemsrespectively.Theseexpertswereknowledgeableaboutresearchintheirownfields.

Theexpertsreceivedalistwiththefollowingitems:Idon'tneedtothinkmuchonhowto
use [technology x] as it seems wellsuited with my old habits; My habits automatically
match with the way I would use [technology x]; I believe that using [technology x]
regularly, would match with my current habits; Working or playing with [Technology x]
goes very well with my habits; Working or playing with [Technology x] fits very well my
habits; Using [technology x] fits very well with my current habits; Beginning to use
[technologyx]frequently,wouldnotrequiremetochangemyhabitsinanuncomfortable
way;Includingtheuseof[technologyx]inmylife,wouldbecompatiblewithmyhabits;I
definitelyfeelcompatibilitybetweenmyhabitsandusing[technologyx];andMyhabitsare
attunedwithusing[technologyx].

Each expert got the definition of all the constructs and a summary of the expected
characteristics of the items. They were asked to provide any general or specific feedback.
They were asked to add, eliminate, rewrite, or correct any item. The researcher held
personalmeetings,phonecallsandemailcorrespondencewiththeexpertsuntilafinallist
wasready.Table3.1QSoringOpenExerciseSimilarityMatrix)andFigure1.1Research
Model Simplified) show seven items that were selected as a result of the expert
consultation.Thosesevenitemswereincludedinthenextphase,QSortingexercise.

3.9.3 QSortingExercise

The QSort method is a method of assessing reliability and construct validity of

questionnaire items. It is an empirical approach to evaluate validity during the scale
102

development process (Nahm et al. 2002). Qmethodology and Qsorting was introduced
originally by Stephenson (1935). The method draws individual subjective standpoints in
modelsthatcanbecomparedtootherindividualsmodels(Stephenson1988).Thomasand
Watson(2002)arguethatintheinformationsystemsfield,thismethodoffersasystematic
andrigorousquantitativemeansforexamininghumansubjectivity.Qmakesreferenceto
datacontainingrespondentssubjectivity,asopposedtoRwhichdenotesdatacontaining
objectivemeasures(Brown1997).

Qmethodologyhasbeenusedinavarietyofwaysandwithdiversepurposes,oneofthem,
scaledevelopment.Whiledevelopingascale,theobjectiveofthisexerciseistoverifythe
convergent and discriminant validity of the scales by examining how the items are sorted
intovariousconstructcategories(Moore&Benbasat1991).Thismethodwasusedbyoneof
theseminalauthorsinInformationSystems,FredDavis(Davis1989).Davisusedthismethod
tovalidatetheconstructsintheTechnologyAcceptanceModel(TAM).

Participants were approached through an invitation published on a local classified ads

websiteinthecityofMelbourne,Australia.ThecallwasalsopublishedinRMITUniversity
and Facebook ads. Fortyfive adults volunteered to participate in this research. The
participants received a free lunch and a small gift in return for their participation. This
method does not require a large number of participants. Davis (1989) conducted a card
sorting exercise (Q) with only 15 respondents. However, Thomas and Watson (2002)
recommend30to60participants.

The volunteers were asked to attend to one of three sessions in RMIT University. A
classroom with computers was booked in advance. The researcher used Optimal Sort
online software to conduct the QSorting exercise (optimalworkshop.com, version: N/A).
Optimal Product Ltd trading as Optimal Workshop provided a free one month license in
supporttothisresearch.Duringthesessiontheparticipantscompletedtwoexercises:open
andclosedQSorting.

103

3.9.4 Opensortexercise

Firstly,anopensortingexercisewasconducted.WithinOptimalSort,theparticipantsfound
34 virtual cards corresponding to seven constructs, randomly presented. Each card
contained a sentence (measurement item). Participants were asked to sort the cards by
conceptual similarity in as many groups they considered appropriate and name their
categories.Exceptfortechnicaldoubtsonhowtodraganddropcards,orhowtofinishthe
exercise,nootherhelpwasprovidedtoavoidbias.

Theparticipantswerenotrestrictedaboutthenumberofgroupshe/shecouldmake.There
was no minimum or maximum limit for the number of virtual cards that could be put
together. However, each card could be placed in only one group, and a valid response
requiredatleasttogroupsofcards.Theresearcheremphasizedthatthesimilaritiesshould
bebasedonunderlyingconceptsandnootherpatterns.

Once the participants were happy with their responses, they submitted by clicking on a
finish button. The exercise took the participants 17.2 minutes on average, and 43
participantssubmittedvalidresults.

3.9.5 Closedsortexercise

Secondly, a closed sorting followed the open exercise in the same session. The same 34
cards appeared randomly placed on the participants screens as in the open exercise.
However, this time the name of the construct was the heading of each category. The
participantswereaskedtoreadthetitles,andtoallocatethecardsinthegroupwherethey
betterfitted.Iftheyfoundacardnotfittingany,theyweregiventheoptiontoplaceiton
othercategory.Theresearcherdrewtheattentionoftherespondentstoconsiderhowwell
thestatementsreflectthegivenconcept.

Theclosedexerciseisdifferenttothepreviousexercisebecause,inthiscase,therewerea
definitepredeterminednumberofoptions.Theparticipantshadtoplaceallthecardsinat
least one category to produce a valid result. The closed exercise was completed in 8.3
minutesonaverage,and44peoplesortedall34cards.
104

3.9.6 AnalysesfortheQSortingexercises

OptimalSortsoftwareprovidesvariousanalysesfortheopensortexercise.Similaritymatrix
anddendrogramswereselected.Thesimilaritymatrix(Table3.1)groupsthemeasurement
itemsbypairs.Theresultofeachcellcanrangefrom0to100,whichisthepercentofthe
participantswhoagreethatthosetwoitemsbelongtogether.ThealgorithmofOptimalSort
alsoattemptstoclustersimilarcardsalongtherightedgeofthematrix.Thoseindarkerblue
areclosertoeachotherthantheonesinlighterbluecolor.

The dendrograms analysis (Figure 3.3) is a hierarchical cluster method commonly used in
taxonomy (Waterman & Smith 1978). The dendrograms derive from factor analysis, and
theycanbeusefultographicallyassesstherelativeclassificationstrengthsbetweenfactors.
Items within the same class are more similar to each other than independent items.
Dendrogramscanshowcompactclasses,whentheitemsareverysimilartoeachotheror
isolateditemswhentheyarenot(VanSickle1997).

3.9.7 ResultsfortheQSortingexercises

Table3.1showsthatalltheitemsofUTAUT(Venkateshetal.2003)wereconsideredhighly
similarinmeaningbytherespondents.Thefirstthreeitems(BI3,BI2&BI1intheorderof
OptimalSortsoutput)showthat97%oftheparticipantsplaceditemBI2andBI3together;
93%oftherespondentsagreedthatBI1andBI2conveyedthesameorverysimilarmeaning;
95% of them placed BI1 and BI3 in the same group. Although the items for behavioral
intention (BI) (Venkatesh et al. 2003) and the ones for actual behavior (AB) (Liang et al.
2010) were perceived as very similar, it is possible to see that BI1 and AB2 obtained only
60%oftheagreement,whereastheinternalsimilarityforeachconstructgotbetween86%
97%oftheagreement.Therestofthetablecanbereadinthesamewaybymatchingpairs
ofitems.

OfthesevenHTFitems,thestrongestsimilarityoccurredforHTF2,4,and5.Althoughitem7
isapartfromtheHTFcluster,itisimportanttoremarkthatitwasplacedwithEEitemsas
much as it was placedwith HTF items. Facilitating conditions (FC) obtained relatively low

105

percentageofagreement.BesidesFC1andFC2,nootherpairobtainedmorethan40%of
the agreement. Effort expectancy (EE) items ranged from 67% to 81% of the agreement,
forming a clear and compact cluster. PE14 were the original items of UTAUT and they
ranged between 69% and 93% of the agreement. Except for PE5 and PE9, the additional
itemsforPEalsoachievedveryhighlevelsofagreementamongtheparticipants.Still,inthe
finalanalysistheresearcherstucktotheoriginalitemsasoriginalonesdidnotpresentany
problemastheresearcheranticipated.

QSoringOpenExerciseSimilarityMatrix
BI3- I plan to use iPhone in the next 30 days.
97 BI2 - I intend to use iPhone in the next month.
95 93 BI1 - I predict I would use iPhone in the next 4 weeks.
60 62 60 AB2 - In the last month, I used iPhone frequently.
55 58 53 90 AB1 - In the last month, I spent a lot of time using iPhone.
55 53 60 86 93 AB3 - In the last month, I used iPhone intensively.
6 4 11 11 13 20 HTF2 - Using iPhone fits very well with my habits.
11 13 16 11 11 13 74 HTF5 - I think using iPhone doesn't set me apart from my habits.
6 9 11 11 13 16 72 72 HTF4 - Including the use of iPhone in my life is compatible with my normal behavior.
6 9 9 13 16 13 41 51 53 HTF3R - Using iPhone frequently, requires me to change my habits in an uncomfortable way.
9 11 11 9 9 9 11 20 20 37 FC3 - iPhone is not compatible with other systems I use.
13 13 9 9 9 6 13 16 13 11 37 HTF6 - I tend to use technologies which seem somehow very similar to iPhone.
2 2 0 2 2 2 23 18 23 18 9 34 SI4 - In general, my environment has supported the use of iPhone.
11 11 9 2 2 2 9 11 13 18 9 23 55 SI1 - People who are important to me think that I should use iPhone.
9 9 6 2 2 2 11 16 18 23 9 23 53 95 SI2 - People who influence my behavior think that I should use iPhone.
9 9 6 4 4 4 4 6 6 16 11 18 48 79 74 SI3 - People with some kind of authority in my life, have been helpful in the use of iPhone.
2 4 2 2 2 0 6 13 13 11 16 25 37 37 39 48 FC4 - A specific person (or group) is available for assistance with iPhone difficulties.
4 2 2 4 6 6 20 20 27 16 20 30 27 11 13 18 39 FC1 - I have the resources necessary to use iPhone.
4 2 4 6 6 9 20 18 23 18 20 27 20 4 4 13 30 69 FC2 - I have the knowledge necessary to use iPhone.
4 2 6 9 9 13 23 18 25 20 25 23 18 2 2 16 34 46 65 EE4 - Learning to operate iPhone has been easy for me.
6 4 6 11 9 11 25 20 30 16 27 20 20 2 2 11 30 48 62 81 EE3 - I have found iPhone easy to use.
6 4 6 11 9 11 37 27 32 18 18 20 20 4 4 11 18 39 53 72 76 EE1 - My interaction with iPhone has been clear and understandable.
4 2 9 6 6 13 23 23 23 16 27 23 11 2 2 9 25 51 62 76 74 67 EE2 - It has been easy for me to become skillful using iPhone.
6 4 11 6 9 16 39 32 48 32 32 25 23 4 6 6 20 41 37 60 55 51 53 HTF1 - I don't need to think much on how to use iPhone as everyth
6 4 9 9 11 16 46 32 41 27 20 32 25 4 4 4 9 30 34 41 37 39 34 62 HTF7 - Working or playing with iPhone goes very well with the
4 4 4 9 9 9 27 18 13 9 11 18 18 4 4 6 4 16 13 6 11 11 13 11 41 PE2 - I find iPhone useful in my job.
4 4 6 9 9 11 32 27 23 11 6 9 9 0 2 0 4 18 11 6 9 13 18 16 34 72 PE1 - Using iPhone improves my performance in the
4 4 6 9 9 11 30 23 20 11 9 9 11 0 0 0 2 16 13 11 11 13 18 16 39 72 93 PE8 - Using iPhone increases my capacity for do
4 4 4 11 9 9 27 18 13 13 6 9 16 6 6 4 4 11 11 4 11 13 11 11 34 72 81 79 PE4 - Using iPhone increases my productivi
9 9 9 16 13 13 27 20 18 11 9 6 9 0 2 2 6 16 9 6 13 13 13 16 30 69 86 79 88 PE3 - Using iPhone enables me to acco
6 6 6 11 9 9 23 20 23 9 9 9 9 4 6 2 6 16 11 4 13 13 11 13 27 65 81 76 79 81 PE7 - Using iPhone can enable me
9 9 9 11 9 9 41 34 27 11 16 11 18 2 2 2 4 16 16 11 18 23 20 16 41 60 60 65 58 58 60 PE6 - I find iPhone useful in at
2 2 2 6 4 4 11 13 9 11 6 6 16 16 18 13 9 13 6 6 9 9 11 13 18 46 46 39 53 48 51 32 PE9 - If I used iPhone, I wo
2 2 4 2 2 4 11 13 11 13 4 6 16 16 18 13 11 13 6 9 6 4 11 16 23 53 51 46 51 46 51 30 90 PE5 - If I use iPhone,

GeneratedwithOptimalSort
BIBehavioralIntention SISocialInfluence
ABActualBehavior EEEffortExpectancy
HTFHabitTechnologyFit PEPerformanceExpectancy
FCFacilitatingConditions
Table3.1QSoringOpenExerciseSimilarityMatrix (Source:Author)
(SeethefulllistofitemsinAppendix6)

Figure3.3QSoringOpenExerciseDendrogramsshowstheresultsoftheOpenExercisein
form of dendrograms. Consistent with the similarity matrix, dendrograms show that
approximately95%oftheparticipantsagreedthattheitemsforbehavioralintention(BI13)
belongtogether,90%establishedthatactualbehavior(AB13)itemsclusteredinthesame
106

 concept.Inthatwaythebestitemsmeasuringtheconstructshereincludedwouldbe:BI13
(95%), AB13 (90%), SI13 (79%), EE14 (76%), HTF2, 4&5 (72%). On the other hand, the
poorest agreement was achieved by facilitating conditions FC14 with only 37% assigning
themtogether.ApartfromtheFCitems,fewotherswereisolatedbelow60%ofagreement:
SI4,HTF3,andPE5b.

QSoringOpenExerciseDendrograms

GeneratedwithOptimalSort
BIBehavioralIntention SISocialInfluence
ABActualBehavior EEEffortExpectancy
HTFHabitTechnologyFit PEPerformanceExpectancy
FCFacilitatingConditions
Figure3.3QSoringOpenExerciseDendrograms (Source:Author)
(SeethefulllistofitemsinAppendix6)

Table 3.2 shows the results for the closed exercise. The analysis software clusters those
itemsthatgothigherpercentagesofrespondentsplacinganiteminacategory.Themain
difference to the open exercise is that the closed exercise provides seven predetermined
categorieslabeledastheconstructsandoneothercategory.

OptimalSort software presents the results in clusters indicated in blue color. Only
percentagesrangingbetween20%and100%areclusteredbythesoftware.Incaseswhere
an item obtains 20% or more in two different categories, OptimalSort marks blue the
highestvalue.

107

Actualbehavior(AB13)clusteredtogetherwithvaluesrangingfrom73%to80%suggesting
highconvergentvalidity.However,twoeffortexpectancy(EE1&3)itemswereclassifiedas
actual behavior by 39% and 30% of the respondents too. Items EE2 and EE4 were mainly
placedundertheconceptofeffortexpectancy,buttheyonlywereplacedinthiscategoryby
27%and34%oftherespondentsrespectively.FacilitatingconditionsitemsFC1,FC2andFC4
wassortedinthiscategory52%,39%and59%ofthetimes.ItemFC3wasclassifiedashabit
technologyfit27%,andonlyin18%ofthetimesasitsowncategory.

All the items corresponding to the newly developed scale, habittechnology fit, were
clustered together by OptimalSort software. Items HTF2, HTF3, HTF4, HTF5 and HTF7
obtainedthehighestpercentages(57%to80%).OnlyHTF1andHTF6obtainedlessthan50%
oftherespondentsclassifyingitinothercategorythanitsown.

Theoriginalitemsforperformanceexpectancy(Venkateshetal.2003)withthebestcontent
validitywerePE4(73%),PE1(68%)andPE3(61%).Theitemwiththelowestplacementinits
correspondingcategorywasPE2(45%).Otheritemsforperformanceexpectancynotfrom
the original measurement achieved high percentage of placement in this category. The
itemsPE5andPE8achieved61%and59%.

Socialinfluenceitemsweretheoneswiththehighestlevelsofagreement.ItemsSI1andSI2
were placed in this category by 98% of the respondents, SI3 by 91%. It contrasts with SI4
whichwassortedassocialinfluenceonly48%ofthetime.

108

 QSoringClosedExercisePopularPlacementMatrix

Actual Effort Facilitating Habits- Performance Social Behavioral
Behavior Expectancy Conditions Technology Expectancy influence Intention Other
(AB) (EE) (FC) Fit (HTF) (PE) (SI) (BI)
AB2 80% 2% 5% 2% 5% 7%
AB3 75% 5% 2% 11% 7%
AB1 73% 2% 2% 7% 7% 9%
EE3 39% 32% 14% 11% 2% 2%
EE1 30% 23% 20% 11% 2% 7% 7%
EE4 18% 34% 11% 23% 2% 7% 5%
EE2 20% 27% 14% 25% 7% 2% 5%
FC4 2% 7% 59% 5% 18% 2% 7%
FC1 9% 14% 52% 7% 2% 5% 11%
FC2 20% 18% 39% 11% 2% 2% 7%
HTF2 7% 5% 86% 2%
HTF5 5% 5% 2% 77% 2% 2% 7%
HTF3 2% 9% 2% 75% 2% 2% 7%
HTF4 7% 5% 14% 73% 2%
HTF7 7% 11% 11% 57% 2% 7% 5%
HTF1 7% 23% 9% 45% 2% 2% 11%
HTF6 5% 2% 7% 43% 5% 20% 18%
FC3 7% 18% 27% 11% 9% 27%
PE4 9% 5% 5% 5% 73% 5%
PE1 16% 5% 5% 2% 68% 2% 2%
PE8 9% 14% 5% 5% 61% 2% 5%
PE3 16% 9% 2% 5% 61% 7%
PE5 14% 5% 2% 59% 9% 11%
PE5b 7% 7% 2% 57% 5% 9% 14%
PE2 27% 5% 11% 7% 45% 5%
PE7 14% 25% 7% 9% 36% 9%
PE6 23% 9% 9% 16% 32% 2% 7% 2%
SI1 2% 98%
SI2 98% 2%
SI3 5% 2% 2% 91%
SI4 43% 7% 48% 2%
BI2 9% 2% 5% 77% 7%
BI3 9% 2% 2% 2% 2% 73% 9%
BI1 7% 2% 2% 5% 2% 73% 9%

GeneratedwithOptimalSort
(Source:
Table3.2QSoringClosedExercisePopularPlacementMatrix
Author)

Behavioralintentionitemswereconsistentlyplacedinthecorrespondingcategory.BI2was
sortedinthiscategory77%,BI1andBI373%ofthetimes.Theclosedexerciseshowsthat
respondents clearly differentiate between behavioral intention (BI) and actual behavior
(AB). BI items were sorted as AB in less than 10% of the responses. AB3 was placed 11%
underthebehavioralintentionlabel,beingthehighestpercentageofalltheABitemssorted
asBI.
109

Mostoftheitemswererarelysortedintheothercategory.However,HTF6andHTF1call
theattentionastheywereclassifiedasother27%and18%ofthetimes.Inparticular,HTF6
was also sorted by 27% of the respondents into its own category. This indicates relatively
low agreement about the item belonging to HTF, but more critically to the same extent
representingotherunintendedconcept.

3.10 Analysistechniques

Thereareaplethoraofquantitativetechniquesofanalysis,whichcanbeorganizedinfour
categories: description, association, causation and inference. Descriptive techniques are
mainlyusedtoreportthedistributionofasampleacrossarangeofvariables.Theyinclude
measuresoffrequency,centraltendencyanddispersion.Associationtechniquesareusedto
determine the degree to which two variables covary. These include simple correlation,
analysisofvarianceandcovariance,andsimple,partialandmultipleregressions.Inorderto
establish causation, factor analysis, path analysis, structural equation modelling and
regression(simple,partialandmultiple)arecommonlyused.Inferencetechniquesserveto
make estimates of a population from a sample, and to establish if differences or
relationshipswithinasamplecanbeexpectedtooccurotherthanbychance(significance
tests)(Blaikie2010;Tabachnick&Fidell2007).Thisthesismakesuseofthefourcategories
at different moments of the analysis. However, due to the theory testing nature of the
researchquestions,hypothesesandobjectivesofthisthesis,causationanalysisisrequired.

3.10.1 Structuralequationmodelling

StructuralequationmodellingusuallyreferredasSEMhastwosubtechniques.Thefirstone
is variancebased SEM, also known as PLSSEM or simply PLS (partial least squares). The
secondoneisthecovariancebasedSEMwhichisusuallyreferredasCBSEMorsimplySEM.
Variancebased is a causal modelling technique, and its focus is maximizing the variance
explained of the dependent variable. SEM concentrates on estimating the statistical
difference between the structure of theoretical relationships and the data (Hair, Ringle &
Sarstedt2011).

110

The advantages of PLS are it minimizes the residual variances of the dependent variables,
compared with SEM it presents less issues with model identification; it can work with
smaller samples, and can directly incorporate reflective as well as formative constructs.
DisadvantagesofPLSarethattheissueswiththemeasurementmodelhavetobeaddressed
before in order to produce valid results , and it is restricted in theory testing because it
cannotgloballyestimatemodelgoodnessoffit(Hair,Ringle&Sarstedt2011).

SEMhastheadvantageofbeingaconfirmatorytechniqueappropriatefortheorytestingas
itprovidesglobalestimatesofmodelfit,itcanbeappliedtoconfirmatoryfactoranalysisas
well as causal modeling, it can embrace multiple dependent variables and it can estimate
error terms. Disadvantages of SEM are that it requires larger samples (no less than 60
observances),anditsassumptionscanbeveryrestrictive.Thistechniqueassumesnormality,
linearity,andabsenceofmulticollinearity(Tabachnick&Fidell2007).

The choice between variance and covariance based SEM firstly relies on a philosophic
selection criteria. If the purpose is theory testing and confirmation, SEM is appropriate. If
the goal is prediction and theory development, PLS is rather the recommendation (Hair,
Ringle & Sarstedt 2011). Secondly, the selection has to consider the limitations of each
technique.Thisthesisconcentratesontheorytesting;thereforeSEMwasselected.

In order to minimize the disadvantages of SEM, the researcher has taken the following
measures. In regards to the sample, the appropriate number of participants has to be
calculated (a minimumof 210 for this study), and 251 useful responses were obtained. In
ordertominimizetheimpactofthedisadvantages,nonnormalityconcernswereaddressed
usingbootstrappingsubsampling(Byrne2010);multicollinearityconcernsintheextended
model were tackled with complementary analysis techniques such as redundancy analysis
andftests,andavoidingtheinterpretationoftheindicatorsaffectedbymulticollinearityas
explainedinFornellandLarcker(1981).Additionally,theanalysisofthemodelswasalsorun
inPLStocrossvalidatetheresults(seeAppendix10).

In cases, such as in this research, when data normality is not achieved the options are:
applyingdatatransformations(Tabachnick&Fidell2007)orusingatechniqueknownasthe

111

bootstrap (Byrne 2010). The common advantage of these options is to avoid overlooking
the normality assumption and the possibility of producing invalid results. However, a
disadvantage to the data transformations is that it may be more difficult to interpret the
results(Tabachnick&Fidell2007).Additionalbenefitsforthebootstrappingtechniqueare
that it allows assessing the stability parameter estimates, and reporting more accurate
resultswithrelativelysmallsamples(notextremelysmall).Bootstrappingisautomatic,and
easytoset.However,someofitslimitationsincludethatitcannotmakerepresentativea
sample which is not, and for normally distributed data it would produce more biased
estimatesthanthemaximumlikelihoodmethod(Byrne2010).

3.11 Preteststudyprocedure

Theobjectiveofrunningapreteststudywastoobtainfeedbackaboutthesurveyitself.In
online surveys, this step is particularly important because accessibility, flow and technical
issues are a concern. However, other aspects common with a paper survey were also
considered,suchasspelling,wording,readabilityandansweringlength.

The researcher developed a preliminary online questionnaire in Qualtrics online software,

which included greetings, plain language statements (PLS), informed consent button, two
questionnairescomprisingtheitemsfortheUTAUTconstructsplusHTF.Theresearcherand
his supervisors revised the questionnaire thoroughly to eliminate errors in functionality,
spellingandwording.Thepretestwassenttothevolunteerparticipantsforrevision,each
question provided a check box to indicate a problem with the question and a text box to
provide feedback in the necessary cases. At the end a general feedback text box was
available. Informal discussions were held with the participants of the pretest to obtain
deeperfeedback.

Ten people participated in the pretest study. They were selected by convenience among
postgraduatestudents,universitystaffandITprofessionals.Thepreteststudycontributors
received the online survey with additional features for them to provide feedback. The
surveywasafullyfunctionalsurvey,exactlyasthedesignatedinstrumentforthepilotand

112

main studies. The only difference was it included a check box after each question that
openedatextfieldwhenticked.

Eachparticipantreceivedinstructionsfromtheresearcheronaonetoonebasis.Thenthey
beganrespondingtothequestionnairefollowingthesamekindoflinkusedinthepilotand
main studies. The researcher would sit next to the participant watching expressions of
confusion or problems to follow the surveys instructions. At the end of the survey, the
participants found an additional field to provide feedback on the general survey. Each
session ended with an informal conversation with the researcher in which other concerns
weretakenintoaccount.

3.11.1 Questionnairerefinement

The pretest study uncovered aspects that had not been considered before, and were
modifiedtoimprovetheinstrument.Someofthemostrelevantaspectsthataroseduring
thepretestwere:forcedresponses,excessivenumberofdemographicquestions,lengthof
the introduction, repetition of some questions, progress bar and time to complete. The
followingaresomeofthecommentsabouttheinstrumentandtheactionstheresearcher
took.

One of the respondents suggested not forcing responses by making questions mandatory,
especially for open questions. The researcher deemed it preferable to have full
questionnairesreadytobeanalyzed,asthestructuralequationmodellinginAMOSsoftware
requiresnomissingdata.Threeoftheparticipantscommentedabouttheexcessivenumber
ofdemographicquestions,andtheywerereducedtotheminimumnecessary(age,gender,
country,andeducationlevel).Tenparticipantscomplainedaboutthelengthofthetextin
theintroduction.Theresearchersynthetizedthemostimportantinformationandprovided
the option to display the full PLS by clicking a button. Other matters were identified and
corrected,suchasproblemswiththeprogressbar,videospresentingthetechnologiesfor
evaluationwereveryproblematic,whichwassolvedbyincludingalinktoYouTube.

113

The participants reviewed the questionnaire after modification, and reported a significant
improvement.Thetimetocompletethesurveywasreducedto15minutes,theinstructions
wereclearerandthequestionsdidnotseemsorepetitive.Itwaseasiertounderstandand
providedsimpleoptionsfortechnicalproblemswiththevideo.Afterthesecondreview,the
surveywasapprovedtobeusedinapilotstudy.

3.12 PilotStudy

Assessingthefeasibilityofalargescaledatacollectionisthepurposeofthepilotstudy.It
consistsofdeployingsurveydistributionstrategiesandreachingmembersofapopulationin
the same way as a full scale study would be organized. Through the pilot study, the
researchercanevaluatehisproposedrecruitmentapproach,thereliabilityofthescalesand
uncover potential issues with the analysis techniques, the variability in the outcomes and
anylogisticproblemsthatmayberelevantinalargerscaledatacollection.

Forthisthesis,theresearcherconductedapilotstudywithaconveniencesampleofadults
at least 18 years old, who understand English. An invitation was sent to 43 selected
acquaintances,simulatingtheprocessofaseedintherespondentdrivensamplingmethod
(Heckathorn2002).Theemailcontainedafriendlyinvitationinitsbody,aPDFfileattached
withtheplainlanguagestatement(PLS),andahyperlinktoanonlineregistrationform.

Theregistrationformaskedname,emailandnameofthepersonfromwhomtheinvitation
was directly received. Only the email address was a mandatory field, once the participant
wasregisteredheorshewouldreceiveanemailwithauniquelinktothesurvey.

3.12.1 Resultsofthepilotstudy

The pilot study confirmed the appropriateness of the distribution strategy for the survey
andthereliabilityofinstrumentitself.Noproblemswerefoundwiththeinstrumentorthe
scales.Theprocessofinvitation,registrationandsurveyresponseranslickly.Thereliabilities
foreachconstructwerecalculated,andalltheconstructsobtainedaChronbachsalfaabove
0.7.

114

The only concern that arose from the pilot study was a low likelihood of the data being
normally distributed. This would be an issue for covariancebased structural equation
modeling. However, given the warning arisen by this possibility, the researcher found
alternativesinordertoworkwithnonnormaldata.Thus,theresearcherproceededtothe
mainstudy.

115

3.13 Thefinalinstrument

MeasurementItemsandDemographicQuestions
Class Item Characteristics
LATENTVARIABLES
AB1 Inthelastmonth,Ispentalotoftimeusing[technology].
7pointLikertscale(Strongly
AB2 Inthelastmonth,Iused[technology]frequently.
Disagree/Stronglyagree)
AB3 Inthelastmonth,Iused[technology]intensively.
BI1 Ipredictlwoulduse[technology]inthenext4weeks.
7pointLikertscale(Strongly
BI2 Iintendtouse[technology]inthenextmonth.
Disagree/Stronglyagree)
BI3 lplantouse[technology]inthenext30days.
EE1 Myinteractionwith[technology]hasbeenclearandunderstandable.
EE2 Ithasbeeneasyformetobecomeskillfulusing[technology] 7pointLikertscale(Strongly
EE3 lhavefound[technology]easytouse. Disagree/Stronglyagree)
EE4 Learningtooperate[technology]hasbeeneasyforme.
Ihavetheresourcesnecessarytouse[technology].
FC1
Ihavetheknowledgenecessarytouse[technology].
FC2 7pointLikertscale(Strongly
[Technology]isnotcompatiblewithothersystemsIuse.
FC3 Disagree/Stronglyagree)
A specific person (or group) is available for assistance with [technology]
FC4
difficulties.
I don't need to think much on how to use [technology] as everything I have
beendoinginmylifeissoclosetoit.
HTF1 Using[technology]fitsverywellwithmyhabits.
HTF2 Using [technology] frequently, requires me to change my habits in an
HTF3 uncomfortableway.
7pointLikertscale(Strongly
HTF4 Including the use of [technology] in my life is compatible with my normal
Disagree/Stronglyagree)
HTF5 behavior.
HTF6 Ithinkusing[technology]doesn'tsetmeapartfrommyhabits.
HTF7 ltendtousetechnologieswhichseemsomehowverysimilarto[technology].
Workingorplayingwith[technology]goesverywellwiththewaysIhavelearnt
howtodothings.
PE1 Using[technology]improvesmyperformanceinthecontextwhereIuseit. 7pointLikertscale(Strongly
PE2 Iwouldfind[technology]usefulinmyjob. Disagree/Stronglyagree)
PE3 Using[technology]enablesmetoaccomplishtasksmorequickly.
PE4 Usingwould[technology]increasemyproductivity. (Davis1989;Venkateshetal.2003)
PeoplewhoareimportanttomethinkthatIshoulduse[technology].
SI1
PeoplewhoinfluencemybehaviorthinkthatIshoulduse[technology].
SI2 7pointLikertscale(Strongly
Peoplewithsomekindofauthorityinmylife,havebeenhelpfulintheuseof
SI3 Disagree/Stronglyagree)
[technology].
SI4
Ingeneral,myenvironmenthassupportedtheuseof[technology].
MEASURES
Multipleoption,singleanswer(6months
Ifyouhaveused[technology]inparticular,selectthelengthofyourexperience.
EXP orless,1year,2years,3years,4years,5
(Otherwise,selecttheoption'Ihaveneverused[technology])
yearsormore)
7pointLikertscale(CompletelyNon
VOL Inmycase,theuseof[technology]is:
voluntary/CompletelyVoluntary)
DEMOGRAPHICQUESTIONS
Multipleoption,singleanswer(Lessthan
AGE Age:
24,25to44,45to64,65orover)
Multipleoption,singleanswer(Listof
COU Inwhichcountrydoyoulive?
countries)
Multipleoption,singleanswer(Lessthan
HighSchool,HighSchool/GED,Some
College,2yearCollegeDegree/Associate's
EDU Selectyourhighestcompletedlevelofeducation
Degree/Diploma,4yearCollegeDegree/
BachelorDegree,MastersDegree,Doctoral
Degree)
GEN Gender: Dualoption(Male/Female)
AB Actual behavior, BI Behavioral Intention, EE Effort Expectancy, FC Facilitating Conditions, HTF Habittechnology Fit, PE
PerformanceExpectancy,SISocialInfluence,EXPExperience,VOLVoluntariness.
Table3.3MeasurementItemsandDemographicQuestions (Source:Author)

116

The survey was organized in five sections: plain language statement (PLS) and informed
participation agreement, selection of two technologies and video introductions to them,
questionnaire for the first technology, questionnaire for the second technology, and
demographicquestions.

3.14 Mainstudy

Thenatureofthemainstudyispurelyquantitative.Theinstrumentwasdesignedtobeself
administered by individuals through a website on the Internet. The main study is cross
sectionalgiventhelimitationsoftimeforthisresearchproject.

Sevenhedonicandutilitarianpiecesoftechnologywereincludedinthestudy.Thesewere:
Facebook,GoogleDocs,MicrosoftOffice365,PayPal,Xbox360onlinegaming,ZohoSuite,
Sales Force Cloud. Respondents selected the most familiar and the most unfamiliar
technology,andsurveysfortheseoptionswerealternatedtokeepabalanceinresponses.

In order to test item validity and the research hypotheses in a larger scale, a refined
questionnaire was used to conduct a survey with 251 respondents who usable
questionnaires.

ThequestionnairewascreatedandrefinedanddistributedontheInternetusingQualtrics
(www.qualtrics.com). The survey was opened for three months, and respondents were
encouraged participate and invite other people to respond through email and social
networks.

3.15 Dataanalysisprocedures

The analysis method followed processes of three main blocks or phases. The first part
consisted of preparing data for analysis, cleaning and formatting data. The second phase
consistedoftheexaminationoftheitemsandthefactors.Oncetheresearcherconfirmed
thereliabilityoftheitemsandperformedfactoranalyses,theprocesswenttothelastblock.
Thethirdphaseexaminesthehypothesizedrelationsamongconstructsandcontraststhem
withtheempiricalfindings(Figure3.4).

117

 ANALYSISPROCESSDIAGRAM

Phase1 Phase2 Phase3

Figure3.4AnalysisProcessDiagram (Source:Author)

ThisstudyconductedExploratoryFactorAnalysis(EFA)usingIBMSPSSStatisticsversion21,
64bitedition,followedbyConfirmatoryFactorAnalysisusingIBMSPSSAmos21.0.0(Build
1178). Based on the results of EFA and CFA, The Stats Tools Package version update
13/12/2012 (Gaskin 2012; Hair et al. 2010) for Microsoft Excel and Parallel Analysis using
Oconnors(2000)algorithmforSPSS(seeAppendix8)aidedtheassessmentofDiscriminant
andConvergentValidity.

In practice, many research procedures are iterative rather than linear. However, reports
holdalinearlogic.Inordertomaketheresultsclearer,theywillbepresentedgroupedby
procedure,asshowninFigure3.4,andnotstrictlyinchronologicalordertoavoidrepetition
andconfusion.However,Section3.16presentsadetailedflowchartinchronologicalorder
where the original plan is contrasted with the deviations (Figure 3.5 Planned vs Real
AnalysisFlow).

118

3.15.1 Datapreparation(Phase1)

The first step in the process was preparing data for analyses. The assessment included
handling missing data, outliers and testing for the assumptions of multivariate analysis
(normality, homoscedasticity, linearity and multicollinearity). Concerns with normality
arose,andweretreatedduringthetestsofthemeasurementandstructuralmodelsguided
byapplyingtherecommendationperformingbootstrappingsubsampling(Byrne2010).

3.15.2 Reliabilitytest

Reliabilityistheextenttowhichavariableorgroupofvariablesisconsistentinwhatthey
intend to measure. In contrast with validity, reliability is not related to what should be
measured, but to how it is measured. Reliability is the degree to which the observed
variable measures the true value free from error. A more reliable measure is one that
consistently behaves in the same way after repeated measurements. Although reliability
and validity are different concepts, reliability is an indicator of convergent validity.
Therefore, in the process of research literature recommends assessing variables and their
measurementinordertochoosethehigherreliability(Hairetal.2010).

Intheprocessofassessmentofthemeasurementitems,thisstudytestedreliabilitybytwo
methods, Cronbach Alpha and Composite Reliability. Cronbach Alpha is a reliability
coefficient which evaluates a complete scale. Literature commonly agrees that a reliable
scale should achieve at least .70 (Cronbach 1951; Hair et al. 2010; Robinson, Shaver &
Wrightsman 1991). The same applicable .70 threshold is the recommendation for
Composite Reliability (coefficient omega). In exploratory research, .60 is the minimum
acceptable value (Bacon, Sauer & Young 1995; Hair et al. 2010). The value obtained by
facilitatingconditionswerelowerthatthecommonlyacceptable,andcouldnotbesolvedby
droppingindividualitems(facilitatingconditions=.505,=.59).Theconstructwaskept
at this stage, but dropped completely after not meeting the minimum requirements in
exploratoryfactoranalysis.

119

3.15.3 ExploratoryFactorAnalysis

Thegoal of exploratoryfactoranalysis(EFA)istoreducealargenumberofmeasurement
itemstoasmallernumberoffactors.Attheend,thistechniqueaimstoprovidereliableand
interpretable factors as an output. Factors are interpreted by the correlations between
variables.Thismethodisexploratoryinnature,anddecisionsaboutthenumberoffactors
andtherotationtypeusuallyarepragmaticratherthantheoryoriented.EFAwasdesigned
for situations where the link between observed and latent variables is unknown (Byrne
2010;Tabachnick&Fidell2007).

ThisstudyconductedExploratoryFactorAnalysisusingIBMSPSSStatisticsversion21,64bit
edition. The analysis included calculation of univariate descriptives, initial solution,
coefficients,determinantandKMOandBartlettstestofsphericity.Themethodselectedto
analyze the correlation matrix was Principal Axis Factoring. The Factor analysis rotation
methodusedwasVarimax(Leech,Barrett&Morgan2011;Tabachnick&Fidell2007).There
were no missing values in the dataset at the moment of the analysis. However, the
researcherselectedlistwisecaseexclusion(seeAppendix8fordetailedsyntaxused).

EFA does not mandate normality, although it is desirable. Linearity is necessary in the
relationship among pairs of variables, and the analysis is degraded when linearity fails. In
such cases, Tabachnick, Fidell & Osterlind (2007) recommends variables transformations.
Detectingandreducingtheinfluenceofunivariateandmultivariateoutliersisimportantin
EFA;noinfluentialoutliersweredetected.However,thedeterminantvalueindicateditem
multicollinearity. Collinear items are redundant, and if used they inflate the size of error
terms. Testing models in the presence of item multicollinearity would weaken the results
(Tabachnick&Fidell2007).Therefore,factorscoreswereusedtosolvethisissue.

Therearetwomainclassesoffactorscore:refinedandnonrefined.Nonrefinedmethods
areadvantageousbecauseofthesimplicityintheircalculation.However,thesetechniques
can be difficult to interpret. Refined methods share the advantage of precision. Between
Bartlett Scores, Regression Scores, and AndersonRubin Scores, the first method was
selected.Thereasonwasitpreservesthevariabilityinrawdata,anditminimizesthesum

120

of squared components for the error factors across the set of variables. This method
delivers factor scores, which highly correlate to their corresponding factor and not with
other factors. An advantage of Bartletts approach over Regression Scores and Anderson
RubinScoresisthattheprocedureproducesunbiasedestimatesofthetruefactorscores
(DiStefano,Zhu&Mindrila2009,p.4),andrepresentsnosignificantdisadvantage.

Based on Hair et al. (2010) and Tabachnick and Fidell (2007) this analysis considered
loadings as small as .35 in the interpretation. However for practical significance, it
disregarded values smaller than .50. Only the items with loadings greater than .50 were
kept, any crossloading equal or higher than .50 was used as a criterion for elimination.
Table 4.4 shows the measurement items reduction to a single composite variableand the
nonreduceditems.

3.15.4 Confirmatoryfactoranalysis

ConfirmatoryFactorAnalysis(CFA)isatheorydrivenconfirmatorytechniqueasindicatedby
its name. The researcher uses a hypostatized model to estimate a population covariance
matrixwhichthealgorithmcompareswiththeobservedcovariancematrix.Schreiberetal.
(2006) explain that it is necessary to have the smallest reachable difference between the
two matrices. Derived from CFA it is then possible to determine convergent and
discriminantvalidityforthemeasurementofaconstruct(Hairetal.2010).

This study used IBM SPSS Amos 21.0.0 (Build 1178) to compute Confirmatory Factor
Analysis.Intheestimationofthediscrepancy,themethodofmaximumlikelihood(ML)was
the selection. Byrne (2010) recommended the following settings: unbiased covariance
supplied as input: unbiased, covariance to be analyzed: Maximum likelihood, 500 random
permutations, Bootstrap: 1000 samples, 90 percentile confidence level, 90 biascorrected
confidenceintervals,andthebootstrapML.

This study used SPSS v.21 to evaluate the assumptions of multivariate normality and
linearity.Datadidnotmeetthenormalityrequirementapriori(seeTable4.5Assessment

121

ofNormality),buttheresearchertookmeasurestoavoidviolatingthecriticalassumptionof
normalityincovariancebasedstructuralequationmodeling(forboth,CFAandSEM).

Another alternative to data transformation and violating the assumption of normality

consists of aprocedureknownasthebootstrap.Bootstrappinghandlesnonnormaldata
througharesamplingprocedure.Thistechniqueassumestheoriginalsamplerepresentsthe
population, and it randomly draws multiple subsamples of the same size as the parent
sample. Literature suggests that bootstrapping is an appropriate approach to handle the
presenceofmultivariatenonnormaldatainSEM(Byrne2010;West,Finch&Curran1995;
Yung&Bentler1996;Zhu1997).

Bootstrappinghascharacteristicswhichfacilitatecopingwithnonnormaldataanddoesnot
obscure the interpretation of results as transformations do. Therefore, bootstrapping was
theselectiontohandletheissueofnormality(Byrne2010;Tabachnick&Fidell2007).

ThisresearchconductedaconfirmatoryfactoranalysisforthreemodelsrelatedtoUTAUT:
thebase,extendedandmodifiedmodel.Figure4.23showsthehypothesizedandmodified
models. Three measurement models are offered because the extended model presented
concernsofdiscriminantvaliditybetweenhabittechnologyfitandperformanceexpectancy.
Although, habittechnology fit and effort expectancy appeared differentiated, there was
alsohighcollinearitybetweenthem.Therefore,afterposthocmodificationCFAwasrunfor
thenewmodeltoo.

3.15.5 IndicatorsofModelFit

Confirmatory Factor Analysis and Structural Equation Modeling share a common set of
indicatorsformodelfit.Thistechniqueprovidessupporttoamodeltothedegreethatthe
fittedpopulationcovariancematrixcorrespondstotheobservedsamplecovariancematrix
(Marsh, Balla & McDonald 1988). It statistically tests the entire model, simultaneously, to
determineitsfitwiththedata(Byrne2010).

A typical approach would reject models if the minimum discrepancy, Chisquare ( ), is

large in relation to the degrees of freedom ( ) (Marsh, Balla & McDonald 1988). A
122

benchmark to evaluate has its base on rules of thumb. However, there are three levels,
whichtheliteraturecommonlyconsidersasappropriate, 2(Byrne2010),
3(Carmines&McIver1981), 5(Wheatonetal.1977).Theminimumdiscrepancy
( ) is usually in association with a probability of getting an obtained value for . This
probability( )assumesthemodeliscorrect,opposedtoassumingthatthenullhypothesis
istrue.Therefore,values .05aretherecommendationastheyrepresentthelikelihood
ofgettinga valuebeyondthe valuewhen istrue(Arbuckle2010;Byrne2010).

Browneetal.(1993)endorsetherootmeansquareerrorofapproximation(RMSEA)asone
ofthemostregardedandinformativecriteriatoassessmodelfit.RMSEAdenoteshowwell
the model would fit the population covariance matrix if it were available (Browne et al.
1993).RMSEAisnonstochasticanddoesnotdependonsamplesize.Valueslowerthan.05
indicateagoodfit,between.05and.08representareasonableerrorsapproximation,.08to
.10amarginalfit,andmorethan.10apoorfit.PCLOSEindicatestheprobabilityofRMSEA
tobegoodinthepopulation.Literaturerecommends.50astheminimalacceptablevalue
forPCLOSE(Hairetal.2010;Jreskog&Srbom1996)

ThisstudyusestheChisquareratio( )andRMSEAasthemainindicatorsofmodelfit,
given they provide probability information. However, this chapter also reports: the
Standardized Root Mean Square Residual (SRMR), Goodness of fit index (GFI), Adjusted
GoodnessofFitIndex(AGFI),ParsimonyadjustedGoodnessofFit(PGFI),NormedFitIndex
(NFI),ComparativeFitIndex(CFI),andHoelter's'criticalN'.Table4.6andTable4.16provide
benchmarkvaluesfortheseindicators.

123

3.15.6 ValidityAssessment(phase3)

DiscriminantandConvergentValidity:Thisstudytestedthefulllatentvariablemodelusing
AMOS. In order to evaluate convergent validity, the Composite Reliability (CR) should be
larger than .70, CR should be higher than the Average Variance Explained (AVE), and AVE
shouldbegreaterthan.50(Hairetal.2010,p.709).Discriminantvalidityevaluationconsists
of comparing the Average Variance Explained (AVE) to Maximum Shared Variance (MSV)
and to the Average Shared Variance (ASV). For a factor to attain discriminant validity, the
MSVandASVshouldbegreaterthanAVE(Hairetal.2010).Allfactorsinthebasemodelgot
discriminant validity. The extended model presented discriminant concerns because for
performanceexpectancyandhabittechnologyfitAVEresultedsmallerthantheMSV.

Criterionrelated validity: this assessment reflects the association of a scale with some
criterion. Criterionrelated validity is a temporarily neutralterm incontrast with construct
validity. Criterionrelated validity deals with the empirical relationship between two
variables,ratherthancausalrelationships.Correlationcoefficienthasbeentraditionallythe
indexforCriterionrelatedvalidity(DeVellis2012).

Criterionrelated validity is commonly confused with construct validity as the former is a

foundation for the latter. Construct validity has a direct concern for the theoretical
relationshipbetweenvariables.Incontrast,criterionrelatedvalidityseeswithneutralityat
the correlations, their direction, and their significance. Criterionrelated validity does not
indicate causality, but causality cannot be claimed if the criterionrelated validity is not
achieved first. Criterionrelated validity only reports the fact that variables behave as
expectedinrelationtoothervariables(DeVellis2012).

124

3.16 Analysisprocessoverview:plannedvsreal

Thissectionpresentsanoverviewoftheanalysisprocesscontrastingtheinitialplanandthe
realexecution.Figure3.5PlannedvsRealAnalysisFlow)showsthecomplexityofthereal
processagainsttheplan.

Datapreparation,reliabilitytest,andexploratoryfactoranalysiswereconductedaccording
totheplan.However,onescaleachievedfairlylowreliabilitylevels(facilitatingconditions
=.505,=.59).Droppingitemswasnotenoughtosolvetheissue;stillthefactorwaskept
untiltestedwithexploratoryfactoranalysis.Exploratoryfactoranalysisrevealedconcernsof
convergenceanddifferentiationforfacilitatingconditions.Thereforeithadtobedropped.

Dropping an important factor from the original model (UTAUT), frustrated the plan of
analyzing the unmodified full model before assessing the impact of including the new
construct(habittechnologyfit).Insteadoforiginalmodel,itwasrenamedbasemodelin
order to make clear the only modification to the model was dropping this factor and
consequently its relationships with the moderators of UTAUT. The newlynamed base
modelcomprises:performanceexpectancy,effortexpectancy,socialinfluence,behavioral
intention and actual behavior. The direct and moderated relationships (age, experience,
genderandvoluntariness)remainedunmodifiedfromtheoriginalUTAUTmodel.

Confirmatoryfactoranalysis(CFA)wasrunforthebasemodel.Noissuesemerged,andthe
analysis continued to structural equation modelling (covariancebased = SEM, variance
based=PLS).ThemainanalysistechniqueconductedwasSEM.Butinordertocrossvalidate
results, PLS was run in parallel. The results from SEM informed the findings from
Hypotheses2and3(NotethatH2&H3didnottakethefirstnumberofhypothesesbecause
they are a test of previous theory. The first priority was reserved for the main original
contributionsofthisthesisH1&H2).

CFAwasrunoncemorefortheextendedmodel.Surprisingly,anewresultemerged.Habit
technologyfitandperformanceexpectancy(thenewconstructandthemaindeterminantof
behavioral intention in UTAUT) raised discriminant validity concerns. Regular procedure

125

doesnotallowthecontinuationtoSEMwithoutconvergentanddiscriminantvalidationof
thefactors.Inthosecasestheresearcherisadvisedtocombineconstructs,exceptwhenit
wouldnotmakesenseconceptually.Insuchcases,thesuggestedoptionisdroppingoneof
the constructs (Farrell 2010). Since the Qsorting exercises provided strong empirical
evidenceontheconceptualdifferenceoftheseconstructs,theresearcherfacedaconflict:
droppinganoldstableconstructordroppingtheoneonwhichthisthesisisfocused.

FornellandLarcker(1981)demonstratedavalidoptionthatproducesvalidresultsfor:the
effect of independent variables upon dependent variable (redundancy test), statistical
significanceoftherelationshipsbetweenindependentanddependentvariableinstructural
models(ftest),and,surprisingly,modelfit(SEM).However,interpretationofthecorrelation
coefficientsorlineardirectionalityisnotallowedinthisoption.Theseresultsinformedthe
findings from Hypothesis 1 and 5. But, the alternative did not provide an outcome for
Hypothesis1a,andonlyapartialresultforHypothesis4(H4requireslineardirectionalityto
evaluatecriterionrelatedvalidity).

It was initially contemplated as a process for planning the posthoc modification of the
extendedmodel.Inthemodificationplanningstage,theresultsandtheliteraturewouldbe
contrastedinordertoconductatheoryguidedmodificationofthemodel.Theunexpected
use of a redundancy test provided new information about the relative value of habit
technology fit against performance expectancy and effort expectancy. Based on Byrne
(2002) and Polkowski (2013), a Venn diagram showed that habittechnology fit contained
the effect size of the other two constructs almost completely, plus an additional margin.
Droppinghabittechnologyfitwouldmeanlosing5.2%oftheeffectsize,whereasexcluding
performanceexpectancyandeffortexpectancytogetherwouldsignifyonlyalossof0.4%.
Thisresultdoesnotdenytheimportanceofanyoftheconstructs;itjustrevealstheyare
redundant,andanewmodelcanbeproposedwithoutlosingexplanatorypower.

126

 PLANNEDVSREALANALYSISFLOWCHART

Figure3.5PlannedvsRealAnalysisFlowchart (Source:Author)

CFAwasconductedonthefactorsofthenewlyvisualizedmodel,andnoconcernsemerged.
Then, the new model was specified in SEM (and secondarily in PLS). The new model
achievedoptimalmodelfitandchisquaregotstatisticalsignificance.Thisledtoprovidean
127

answer to Research Question 3. Although limited to the base model, it was possible to
improve UTAUT by including habittechnology fit and conducting posthoc model
modification.

Fromthemodifiedmodel,itbecamepossibletoassessHypothesis1a.Havingovercomethe
collinearityissuescomingwiththediscriminantconcerns,thehypothesizedmoderatorsof
habittechnologyfitwereassessed.Themodifiedmodelalsocrossvalidatedtheftestand
the redundancy test (which provided a probabilistic significance value and a regression
coefficientfortherelationshipofhabittechnologyfitandbehavioralintentionrespectively).
ThetestofHypotheses1and1aledtoanswerResearchQuestion1.

Finally,notonlyHypotheses2to5providedananswerabouttheeffectofincludinghabit
technology fit in UTAUT, the process itself provided new valuable unexpected findings to
answerResearchQuestion2.

In the next section the decisions of each used technique are discussed, such as problems
andsolutions,softwareemployedandtheoptionsselected.

3.16.1 Analysissoftware

This research used IBM SPSS Amos 21.0.0 (Build 1178) to test the covariancebased
structural equation models. In the estimation of the discrepancy, the researcher selected
thefollowingoptions(Byrne2010):methodmaximumlikelihood,covariancessuppliedas
inputunbiased,covariancesselectedtobeanalyzedMaximumlikelihood,500random
permutations, bootstrap performance 1000 samples, 90 percentile confidence intervals,
90biascorrectedconfidenceintervals,andbootstrapML(MaximumLikelihood).

ForUTAUT,Venkatesh(2003)usedPLSGraphv.2.91.03.04.Despitetheresearchersefforts
tocorrespondwithPLSGraphsauthor,suchversionwasnotobtained.Instead,thisstudy
usedPLSGraph3.0totestthebasemodelwithmoderators,andSmartPLS2.0totestallsix
themodelsasasecondarymethodforcrossvalidation.

128

3.16.2 Issuesandconcerns

A concern arose about the use of SEM in relation with the moderators (Gender, Age,
Experience and Voluntariness). This concern is the categorical nature of gender as a
variable. The concern with gender arises from the impossibility of achieving a normal
distribution.Thisconcernhasastraightforwardsolution.Categoricalvariables,particularly
dichotomous variables such as gender were given a valueof 1 and 2 for male and female
respectively.InaPearsonianspirit,thecategoricalvariablesareviewedasmanifestationsof
continuous normal variables (Muthn 1984, p. 131). This implies that it is appropriate to
usecategoricalvariablesinthiswayinSEM(Arbuckle2010).

By crossvalidating SEM and PLS, other concerns arose when a significant discrepancy
appearedbetweenSmartPLS2.0andAMOS21coefficientsofdetermination(R2andSMC
respectively).TheinflatedcoefficientofdeterminationinAMOSwas.72,whereasSmartPLS
2.0showed.54.Themodelfitindicatorsshowedaremarkablemodelmisfit( >35).
However, this issue found an explanation and solution in the literature. According to
Iacobucci(2010)itisacommonerrortotreatdatainthesamewayinPLSandSEM.Inthe
first,themoderatingvariableismultiplieddirectlyandspecifiedas:

Although the previous specification is suitable for PLS, it creates problems in SEM. Direct
multiplicationofthefactorshighlyinflatesthesquaredMultipleCorrelations(coefficientof
determination).Besides,itdropsthefitofthemodeltounacceptablelevels,increasingthe
possibility of error type II (failing to reject the null hypothesis). Iacobucci (2010) offers an
appropriate alternative, used in this thesis, to test interaction among variables which is
compatiblewithcovariancebasedSEM:

A direct determinant (such as Performance Expectancy) and a moderator (like

Gender)arespecifiedinthestructuralmodelasdeterminantsof .Insteadofspecifyinga
direct multiplication as the third determinant of (See Figure 3.6), the mean
deviationsofbothvariablesaremultiplied .

129

 MODERATORSINSEM

Figure3.6ModeratorsinSEM (Iacobucci2010)

Eventhoughthisreportpresentsinteractionsinasimplifiedform(EXPxSI),theyrepresent

the accurate expression: . This thesis also considered the

implicationsforresultsinterpretationwhenthistransformationisapplied.

According to Whisman & McClelland (2005) other common error in structural equation
modeling(SEM),occursinfailingtoincludetheproductoftheinteractingvariables,aswell
asbothindividualcomponentsinthefollowingway:

Example1:

Inthepreviousexample,behavioralintention(BI)isdeterminedbyperformanceexpectancy
(PE)andtheirrelationismoderatedbytheinteractionwithgender(GEN).Thecorrectway,
followedinthisthesis,specifiestheequationasfollows(Whisman&McClelland2005):

Example2:

3.17 Summary

This chapter justified and explained the methodological decisions made in this research.
Quantitativemethodswereselected,associatedwiththepostpositivismresearchparadigm
of this thesis. Accordingly, a research design was developed. This research is guided by a
deductiveapproachofhypothesistesting.Theapproachtosamplingwaslimitedtoanon
probability sample, the unit of analysis was defined of adults who are Softwareasa

130

Service users in public clouds, and can understand English. The study was set as a cross
sectionaldesignforquantitativedataanalysis.

Theprocedureconsistedoffivestages.Thefirststagebeganwiththeliteraturereview of
habit, personenvironment fit and technology acceptance. It included identifying research
problem, the main theoretical models; developing of a conceptual framework, research
questions, and hypotheses. The second stage included measurement, questionnaire and
sampleframedevelopment.Then,apreteststudywasconductedtorefinetheinstrument,
followedbyapilotstudy.Thethirdstageconsistedinthemainstudy(onlinesurvey).The
fourthwasdataanalyses,andthefifthinterpretationandreporting.

The procedure for measurement development was detailed. This process included: item
generation, expert consultation, and Qsorting exercises (open and closed) to support the
validityofthemeasurementitems.

SEM was defined as the primary data analysis technique. However, it was crossvalidated
with PLS. For Hypothesis 4, multicollinearity concerns emerged. To address these issues,
additionalredundancyofthevarianceandftestwereselected.

131

 CHAPTER4 ANALYSISANDRESULTS

4.1 Objective

Theaimofthischapteristoreporttheresultsoftheanalysisofthedatacollectedinthe
mainstudy.Itpresentsresultsforscalereliabilities,exploratoryfactoranalysis,confirmatory
factoranalysis,validityassessment,modeltesting,andmodelmodification.

ORGANIZATIONOFTHERESULTS

Figure4.1OrganizationoftheResults (Source:Author)

132

ThischapterisorganizedasshowninFigure4.1.Eventhoughthischapterfollowsalogical
order, not every element is chronologically ordered. From descriptive statistics to
exploratoryfactoranalysisalltheresultsarecoincidentallychronological.However,inthe
confirmatory factor analysis section, the results of the three models are presented (base,
extended, and modified). The most important benefit is that it is easier to compare the
results of the same category across the models. A detailed chronological order of the
procedurecanbefoundinFigure3.5PlannedvsRealAnalysisFlowchart.

4.2 DescriptiveStatisticsoftheSample

Qualtrics software was the onlinesurvey tool used to distribute the questionnaire and
collect data. The invitation to participate in the study was received and opened by 1,433
people.Theresearcherreceivedusablequestionnairesfrom251respondents.Theresponse
ratewascalculatedasfollows:

251 100
17.51%
1,433

In the study, slightly more female (54.6%) than male (45.4%) participated. Data was
collected in Australia (36.3%), Mexico (37.0%), and another 23 countries (24.5%).
Respondentsincountriesidentifiedasindividualistaccountedfor58.5%ofthesampleand
collectivistfor41.5%.Thesamplealsoshowedatendencytomoreeducatedpeople,35%of
therespondentsreportedafouryeardegreeand42.3%amastersdegreeatthetimeofthe
survey.

Thisreportpresentscountrybylevelofindividualism.Italsoshowseducationasapartof
the sampleprofile. However, further analysis of these variables is not in the scope of this
thesis.Therefore,thisworkdoesnottestthemfurther.

133

 DESCRIPTIVESTATISTICSOFTHESAMPLE
Respondents
Variable Items
Frequency Percent
Male 114 45.4
Gender Female 137 54.6
TOTAL 251 100
24 4 1.6
2544 102 40.6
Age 4564 122 48.6
65 23 9.2
TOTAL 251 100
Australia(Individualist) 91 36.3
Country Mexico(Collectivist) 93 37.0
(http://geert OtherIndividualist 56 22.3
hofstede.com/) OtherCollectivist 11 4.4
TOTAL 252 100
N/A 10 4.0
LessthanHighSchool 3 1.2
HighSchool/GED 7 2.8
SomeCollege 10 4.0
Education 2yearCollegeDegree 9 3.6
4yearBachelorDegree 88 35.0
MastersDegree 105 41.8
DoctoralDegree 19 7.6
TOTAL 251 100
Table4.1DescriptiveStatisticsoftheSample (Source:Author)

4.3 MeasurementReliability

The Cronbach alpha ( ) for all the constructs was above .70, except for facilitating
Condition, which obtained a Cronbach 0.505(coefficient alpha) and Composite
Reliability(coefficientomega) 0.59.Thevalueobtainedforfacilitatingconditionsisnot
acceptable,anditwasthefirstreasontodropthecompletefactorfromthisstudy.

134

TABLEOFRELIABILITIES
Cronbach CompositeReliability
Construct(Factor)
1

ActualBehavior(AB) .949 .95

BehavioralIntention(BI) .953 .95
EffortExpectancy(EE) .922 .93
FacilitatingConditions(FC) .505 .59
HabitTechnologyFit(HTF) .853 .86
PerformanceExpectancy(PE) .902 .91
SocialInfluence(SI) .858 .87
numberofitemsmeasuringtheconstruct
loadingofthe thmeasure
(Bacon,Sauer&Young1995)
Table4.2TableofReliabilities (Developedbytheauthor)

4.4 ExploratoryFactorAnalysis

Itemsforthesameconstructloadedmainlyonasinglefactorwithvaluesgreaterthan.50
andwithoutcrossloadingonotherfactors.50orhigher.Oneitemforsocialinfluence(SI4)
loadedbelowthesetstandardandtheresearcherdroppedit.Thesamehappenedwithtwo
itemsforhabittechnologyfit(HTF1,HTF3).Inthecaseoffacilitatingconditions,twoitems
(FC3,FC4)loadedbelow.35whichisundertheeliminationthreshold.Oneitemofthesame
construct (FC2) crossloaded greater than .50, and it faced exclusion too (see Table 4.3
RotatedFactorMatrix).Theseresultsprovidedasecondreasoninthedecisionofdropping
thecompletefactorfromthestudy.

In order to attain the rotated factor matrix, this analysis selected an orthogonal rotation
(Varimax). Varimax is the most commonly recommended rotation. Varimax simplifies
columnsoftheloadingmatrixratherthantherowsasQuartimaxwoulddo.Equamaxwould
havebeensuitableasitcansimplifycolumnsandrows.However,itwasnottheselection
becauseoferraticbehaviorsreportedinliterature(Tabachnick&Fidell2007).

135

 ROTATEDFACTORMATRIX
Item 1 2 3 4 5 6 7
EE3 0.889
EE4 0.888

EE2 0.749
EE1 0.671
AB2 0.923
AB3 0.897
AB1 0.829
PE4 0.872
PE3 0.835

PE2 0.692
PE1 0.617
SI2 0.926
SI1 0.762

SI3 0.729
SI4 (0.395)
HTF2 0.626
HTF4 0.639
HTF5 0.545

HTF7 0.518
HTF1 (0.462)
HTF3 (0.381)
BI1 0.712
BI2 0.678
BI3 0.615
FC1 .616
FC2 (.536)*

FC3 (<.35)
FC4 (<.35)
Extraction:PrincipalAxisFactoring ()Droppeditems
Rotation:Varimax/KaiserNormalization *Crossloadingonotherfactors>.5
Table4.3RotatedFactorMatrix (Source:Author)

The1cutoffcriterionfortheEigenvalueindicatedanumberoffactorsequalto6(Leech,
Barrett & Morgan 2011). However a more reliable test, parallel analysis (Oconnor 2000),
determined 8 factors where there should be 7. After removing the facilitating conditions
items,parallelanalysisidentified6factors.Thiscorrespondedtothenumberofconstructs
(SeeAppendix7forresultsandAppendix8forSPSSSyntax).

Avalue>.7inKMOandBartlettstestofsphericityindicatestherearesufficientitemsfor
each factor (Leech, Barrett & Morgan 2011). Before removing facilitating conditions KMO
was 0.926. After removing the items of the construct, the impact on this value was
negligible(KMO=0.917).

136

The determinant value indicated multicollinearity (determinant = 2.01x1009).

Multicollinearityconcernsarisewhentwomeasurementitemscorrelateabove.70toeach
other.HighlycollinearitemscannotbeusedSEMbecauseitassumesnomulticollinearity.
Factorscoreswereusedtocreatecompositevariablesinordertosolvethisissue.Bartlett
Scoresmethodwastheselectionbecauseitpreservesthevariabilityinrawdata(DiStefano,
Zhu&Mindrila2009).

Table4.4showsthemeasurementitemsreductiontoasinglecompositevariableandthe
nonreduceditems.PE3andPE4wereseparatevariables,nowtheresearchermadethema
single composite variable. However, PE1 and PE2 remained in their individual form. The
samelogicappliesfortheothervariables.

COMPOSITEANDINDIVIDUALVARIABLES
PERFORMANCE EFFORT SOCIAL HABIT BEHAVIORAL ACTUAL
EXPECTANCY EXPECTANCY INFLUENCE TECHNOLOGY INTENTION BEHAVIOR
(PE) (EE) (SI) FIT(HTF) (BI) (AB)

EE2,EE3, SI2,SI3 HTF2,HTF4 AB1,AB2,

PE3,PE4 BI1,BI2,BI3
EE4 AB3
SI1 HTF5
PE2 E1 HTF7
PE1
Table4.4CompositeandIndividualVariables (Source:Author)

4.5 Confirmatoryfactoranalysis

This study used IBM SPSS Amos 21.0.0 (Build 1178) to compute Confirmatory Factor
Analysis.Intheestimationofthediscrepancy,themethodofmaximumlikelihood(ML)was
the selection. Byrne (2010) recommended the following settings: unbiased covariance
suppliedasinputunbiased,covariancetobeanalyzedMaximumlikelihood,500random
permutations, Bootstrap: 1000 samples, 90 percentile confidence level, 90 biascorrected
confidenceintervals,andthebootstrapML.

This study used SPSS v.21 to evaluate the assumptions of multivariate normality and
linearity. Data did not meet the normality requirement a priori (see Table 4.5), but the

137

researcher took measures to avoid violating the critical assumption of normality in

covariancebasedstructuralequationmodelingbyusingbootstrap(forboth,CFAandSEM).

ASSESSMENTOFNORMALITY
N Skewness Kurtosis
Variable
Statistic Statistic Std.Error Statistic Std.Error
HTF2 503 .241 .109 .833 .217
HTF4 503 .403 .109 .945 .217
HTF5 503 .464 .109 .848 .217
HTF7 503 .463 .109 .602 .217
PE1 503 .565 .109 .447 .217
PE2 503 .226 .109 1.310 .217
PE3 503 .298 .109 .862 .217
PE4 503 .256 .109 .913 .217
E1 503 .510 .109 .616 .217
E2 503 .625 .109 .367 .217
E3 503 .715 .109 .174 .217
E4 503 .788 .109 .009 .217
SI1 503 .669 .109 .753 .217
SI2 503 .775 .109 .528 .217
SI3 503 .738 .109 .641 .217
BI1 503 .490 .109 1.302 .217
BI2 503 .352 .109 1.356 .217
BI3 503 .507 .109 1.280 .217
AB1 503 2.000 .109 3.090 .217
AB2 503 1.574 .109 .957 .217
AB3 503 1.868 .109 2.101 .217
GEN 503 .188 .109 1.972 .217
AGE_1 503 .552 .109 .987 .217
EXP_1 503 1.894 .109 3.137 .217
VOL1 503 1.083 .109 .095 .217
ValidN(listwise) 503
Table4.5AssessmentofNormality (Source:Author)

Bootstrappinghascharacteristicswhichfacilitatecopingwithnonnormaldataanddoesnot
obscure the interpretation of results as transformations do. Therefore, bootstrapping was
theselectiontohandletheissueofnormality(Byrne2010;Tabachnick&Fidell2007).

ThisresearchconductedaconfirmatoryfactoranalysisforthreemodelsrelatedtoUTAUT:
thebase,extendedandmodifiedmodel.Figures4.7.1,2and3showthehypothesizedand
modifiedmodels.Inthefigures,circlesrepresentlatentvariables,andrectanglesrepresent
measuredvariables.AbsenceofalineconnectingvariablesimpliesneitherUTAUTnorthis
studyhypothesizeddirecteffect.

138

4.6 MeasurementModelValidationBaseModel(UTAUT)

UTAUT hypothesized asix factor model, but EFA led to dropping of facilitating conditions.
Therefore, the base model consists of five factors in this test. The model includes the
followingconstructs:actualbehavior,behavioralintention,performanceexpectancy,effort
expectancy, and social influence. Likert scales ranging 1 to 7 served as indicators of the
latentvariables.

Thisanalysisdiscardedtheindependencemodel,whichteststhehypothesisofallvariables
beinguncorrelated.Themeasurementmodelwasevaluated,anditachievedacceptablefit
at <3(seeTable4.6).Thebasemodel,showninFigure4.2,wasacceptable.

4.7 MeasurementModelValidationExtendedModel

The extended model is a six factor model (UTAUT plus HTF), and it includes the following
constructs: actual behavior, behavioral intention, habittechnology fit, performance
expectancy,effortexpectancy,andsocialinfluence.Likertscalesranging1to7workedas
indicatorsofthelatentvariables.

The extended measurement model was evaluated, and it achieved acceptable fit (at
<3)(seeTable4.6).Theextendedmodel,presentedinFigure4.3,wasacceptable.

4.8 MeasurementModelValidationModifiedModel

The modified model is a four factor structure, and it includes the following constructs:
actualbehavior,behavioralintention,habittechnologyfit,andsocialinfluence.Likertscales
ranging1to7measuredlatentvariables.

The modified measurement model was evaluated, and it achieved acceptable fit (at
<3)(seeTable4.6).Themodifiedmodel,showninFigure4.4,wasacceptable.

139

 MEASUREMENTMODELFITCOMPARISON
Base Extended Modified Benchmark
Indicator
Model Model Model (Byrne2010)
2.387 2.468 2.356 2
0.000 0.000 0.000 p0.05
SRMR 0.025 0.024 0.028 SRMR0.05
GFI 0.963 0.969 0.951 GFI0.90
NFI 0.977 0.982 0.970 NFI0.90
CFI 0.987 0.989 0.983 CFI0.90
RMSEA0.05goodfit;
0.05<RMSEA<0.08acceptablefit;
RMSEA 0.053 0.054 0.052
0.08<RMSEA<0.10marginalfit;
RMSEA0.10poorfit.
PCLOSE 0.342 0.295 0.348 PCLOSE>0.50
HOELTER.05 281 286 269 HOELTER(.05)200
HOELTER.01 315 328 295 HOELTER(.01)200
Table4.6MeasurementModelFitComparison (Source:Author)

140

CONFIRMATORYFACTORANALYSISBASEMODEL

CR AVE MSV ASV AB BI PE EE SI

AB 0.952 0.869 0.524 0.236 0.932
BI 0.954 0.874 0.524 0.343 0.724 0.935
PE 0.864 0.680 0.283 0.233 0.312 0.528 0.824
EE 0.872 0.775 0.283 0.209 0.379 0.500 0.532 0.880
SI 0.880 0.785 0.318 0.232 0.422 0.564 0.523 0.398 0.886
1. Allcorrelationsweresignificant,pvalue<0.001.
2. AB:ActualBehavior;BI:BehavioralIntention;PE:PerformanceExpectancy;EE:EffortExpectancy;SI:Social
Influence.
3. Diagonal elements are the square root of Average Variance Extracted (AVE). The other values are the
correlationsbetweenlatentvariables.
Figure4.2ConfirmatoryFactorAnalysisBaseModel (Source:Author)

141

 CONFIRMATORYFACTORANALYSISEXTENDEDMODEL

CR AVE MSV ASV AB BI HTF PE EE SI

AB 0.952 0.869 0.524 0.226 0.932
BI 0.954 0.874 0.524 0.350 0.724 0.935
HTF 0.828 0.620 [0.709] 0.409 0.427 0.613 [0.787]
PE 0.836 0.633 [0.709] 0.337 0.312 0.526 0.842 [0.795]
EE 0.867 0.766 0.573 0.295 0.386 0.504 0.757 0.579 0.875
SI 0.880 0.785 0.319 0.227 0.423 0.565 0.454 0.517 0.403 0.886
1. Allcorrelationsweresignificant,pvalue<0.001;[] indicatediscriminantvalidityissue.
2. AB:ActualBehavior;BI:BehavioralIntention;HTF:HabitTechnologyFit;PE:PerformanceExpectancy;EE:
EffortExpectancy;SI:SocialInfluence.
3. Diagonal elements are the square root of Average Variance Extracted (AVE). The other values are the
correlationsbetweenlatentvariables.
Figure4.3ConfirmatoryFactorAnalysisExtendedModel (Source:Author)

142

CONFIRMATORYFACTORANALYSISMODIFIEDMODEL

CR AVE MSV ASV AB BI HTF SI

AB 0.952 0.869 0.524 0.292 0.932
BI 0.954 0.874 0.524 0.402 0.724 0.935
HTF 0.836 0.634 0.362 0.243 0.418 0.602 0.796
SI 0.880 0.785 0.318 0.229 0.422 0.564 0.438 0.886
1. All correlations were significant, p value < 0.001; [] indicate discriminant validity issue. 2. AB: Actual
Behavior; BI: Behavioral Intention; HTF: HabitTechnology Fit; PE: Performance Expectancy; EE: Effort
Expectancy;SI:SocialInfluence.3.DiagonalelementsarethesquarerootofAverageVarianceExtracted(AVE).
Theothervaluesarethecorrelationsbetweenlatentvariables.
Figure4.4ConfirmatoryFactorAnalysisModifiedModel (Source:Author)

143

4.9 ValidityAssessmentDiscriminantandConvergentValidity

Results from a first test in EFA showed convergent and discriminant validity for all
constructs, except for facilitating conditions (see Section 4.4 ). Therefore, facilitating
conditionswasnotconsideredanymoreinCFA.Asecond,morerigoroustestwasusedwith
the CFA output. This second test showed a discrimination concern between habit
technologyfitandperformanceexpectancy.Thisindicateshighcorrelationbetweenthetwo
variables,andshouldnotbeinterpretedwithoutconsideringfacevalidity(Hairetal.2010,
p.710).

Table4.6showsthecriterianecessarytoevaluateconvergentanddiscriminantvalidity.The
valuesforAverageSharedVariance(ASV),AverageVarianceExplained(AVE),andComposite
Reliability(CR)areshowninFigures4.13foreachmodel.Thevaluesrepresentingaconcern
areshowninsquarebrackets.

CRITERIAFORCONVERGENTANDDISCRIMINANTVALIDITYCFA
ForConvergentValidity: ForDiscriminantValidity:
CR>0.7 MSV<AVE
CR>AVE ASV<AVE
AVE>0.5
ASVAverageSharedVariance
AVEAverageVarianceExplained
CRCompositeReliability

MSVMaximumSharedVariance
Table4.7CriteriaforConvergentandDiscriminantValidity (Hairetal.2010,p.709)
CFA

This research performed post hoc model modifications. The modified model includes only
habittechnologyfit(HTF)andsocialinfluence(SI)asdeterminantsofbehavioralintention
(BI) and BI as a determinant of actual behavior (AB). For the modified model, all factors
achieveddiscriminantvalidity.

Having completed a test on reliability, exploratory and confirmatory factor analysis, the
followingsectionstestthehypothesesofthisthesis.

144

4.10 TestHypotheses1and1a

H1: HabitTechnology Fit has a positive impact upon behavioral intention

Supported.

H1a: The relationship of habittechnology fit and behavioral intention will be

moderatedbyage,experienceandgender,suchthattheeffectwillbestronger
forolderandmoreexperiencedmenPartiallySupported.

Along the different stages of the analyses, the relationship of habittechnology fit and
behavioral intention was tested. In all the cases the relationship resulted positive and
significant(p<.001).Thisresultssupporthypothesis1.Acceptabletheoryisusuallyexpected
to achieve a minimum of 95% of probability (p value < 0.05) and preferably more than
99.9%(pvalue<0.001)(White&McBurney2010)(seeTable4.8).

RESULTSHYPOTHESIS1
Correlation
Context Method Significant
Coefficient
Single Simplecorrelation 0.557(R2=.310) p<.001
Single PLS 0.562(R2=.315) P<.001
Single ftest N/A p=1.613x1005
ExtendedModel SEM [0.400] [P<.001]
ModifiedModel SEM 0.440 p<.001

[]=WhenHTFwastestedinthecontextoftheextendedmodelinCBSEM,thecorrelationcoefficientcannotbe
interpretedduetothepresenceofhighcollinearity.
BehavioralIntention(BI),HabitTechnologyFit(HTF).
Table4.8ResultsHypothesis1 (Source:Author)

The moderation effect of age, gender and experience were tested in the extended and
modified models. However, the correlation coefficients in the extended model should not

145

be interpreted due to the presence of collinearity between habittechnology fit,

performance expectancy and effort expectancy. The Modified model offers a valid result
(seeTable4.8),withnoconcernsforhighmulticollinearity(asintheextendedmodel).Age
wasfoundasignificant(p<.05)moderatoroftherelationshipbetweenhabittechnology.In
the same relationship gender resulted not significant (p>.05), and experience significant
(p<.001).Therelationshipofhabittechnologyfitandbehavioralintentionwasstrongerfor
olderandmoreexperiencedindividuals.However,thecorrelationcoefficientinbothcases
wasrelativelysmall(<.20).Theseresultssupportageandexperienceasmoderators,butno
gender.Therefore,Hypothesis1awaspartiallysupported.

RESULTSHYPOTHESIS1a
Context Method AGExHTFBI GENxHTFBI EXPxHTFBI
ModifiedModel SEM 0.17* 0.05 0.11***

[]indicatetheeffectcannotbeinterpretedduetopresenceofinteractionterm(moderator)
*pvalue<.05,**pvalue<.01,***pvalue<.001
ActualBehavior(AB),Age(AGE),BehavioralIntention(BI),Experience(EXP),Gender(GEN),HabitTechnologyFit
(HTF),SocialInfluence(SI),Voluntariness(VOL).
Table4.9ResultsHypothesis1a (Source:Author)

In order to test Hypotheses 2 and 4, the researcher tested three modelseach one with
moderators, and then without moderators. UTAUTs full specification as presented in
(Venkateshetal.2003) couldnotbetestedbecausefacilitatingconditionsfactordidnot
achieve convergent nor discriminant validity in the context of this study. Consequently, it
wasnotincludedinanyofthesubsequenttests.

146

 4.11 TestHypothesis2

H2: The original models internal hypotheses will achieve criterionrelated

validitypartiallysupported.

Every model has been tested with and without moderators. The reason is that it is not
possible to give interpretation to the direct relationships in the presence of moderators.
Therefore, the evaluation has to be done in two steps, first with and then without
moderators(Whisman&McClelland2005)(seeFigure4.5BaseModel:UTAUT).

BASEMODEL:UTAUT

x df 2.784, p 0.000, GFI 0.974, CFI 0.984, 3.388, 0.000, GFI 0.782, CFI 0.874,
RMSEA 0.06, PCLOSE 0.171. RMSEA 0.069, PCLOSE 0.000.
Correlations:PEEE=0.529***,PESI=0.523***, Correlations:PEEE=0.531***,PESI=0.523***,EESI
EESI=0.397*** =0.398***
[]indicatetheeffectcannotbeinterpretedduetopresenceofinteractionterm(moderator)
*pvalue<.05,**pvalue<.01,***pvalue<.001
ActualBehavior(AB),BehavioralIntention(BI),EffortExpectancy(EE),PerformanceExpectancy(PE),SocialInfluence(SI).
Modelspecificationbasedon(Venkateshetal.2003)
Figure4.5BaseModel:UTAUT (Source:Author)

The original model could not be tested because facilitating conditions had to be dropped.
UTAUT without facilitating conditions constituted the base model in replacement to the
original model. The base model without moderators presented no deviation from the
hypothesizedcriteria.Performanceexpectancy,effortexpectancyandsocialinfluenceheld

147

a positive relation with behavioral intention. However, age, gender, experience and
voluntarinessfailedtomeetthecriteriainmostoftherelationshipstheywereexpectedto
moderate.Therefore,Hypothesis2waspartiallysupported.

Criterionrelatedvalidityisanecessaryfoundationofconstructvalidity.Itrequiresascaleto
have an empirical association with some criterion (DeVellis 2012). Hypothesis 2 tests the
hypotheses of UTAUT. There were two criteria in this teststatistical significance of the
relationships at least at a p value < 0.05, and direction of the correlation (positive or
negative). In the case of the moderation, the interpretation of positive or negative
correlation should support UTAUTs hypotheses, in this thesis called UTAUTs internal
hypotheses. For example, negative correlation coefficient for gender as a moderator of
performance expectancy and behavioral intention is interpreted as the effect of
performanceexpectancyisstrongerformen.However,theinterpretationdependsonthe
waygenderwascodified.Inthisresearchmaleswerecodedas1andfemaleswerecodedas
2.Therefore,anegativecorrelationisinterpretedasreferringtomen.

Thebasemodelwithoutmoderatorspresentednodeviationfromthehypothesizedcriteria.
Performanceexpectancy,effortexpectancyandsocialinfluenceheldapositiverelationwith
behavioralintention(seeTable4.10).

CRITERIAVALIDATIONBASEMODEL
(WITHOUTMODERATORS)
ExpectedCriteria Outcome
Relationintheory
VariableX Stronger Stronger
XBI Sig Dir Sig Dir
Effect Effect
PE Direct Yes (+) N/A True True N/A
EE Direct Yes (+) N/A True True N/A
SI Direct Yes (+) N/A True True N/A
FC DroppedFactor
Table4.10CriteriaValidationBaseModel(withoutModerators) (Source:Author)

The criteria for the internal hypotheses of the base model were met only for the main
determinantsofbehavioralintention.However,thehypothesizedcriteriawerenotmetfor
theinteractionrelations,exceptforexperiencemoderatingsocialinfluence.

148

Thebasemodel,withmoderators,deviatedfromthehypothesizedcriteria.Allinteractions
betweendirectdeterminantsandmoderatorswereexpectedtobestatisticallysignificantat
leastatapvalue<0.05.However,onlytwointeractionsreturnedastatisticallysignificant
result:genderininteractionwithperformanceexpectancyatpvalue<.05,andexperience
withsocialinfluenceatpvalue<.001.

The direction of the moderation relations was as expected for only three cases. The
theoretical model anticipated that effort expectancy would have a stronger effect on
behavioralintentioninyoungerpeople;socialinfluencewouldhaveastrongereffectonBI
on early stages of experience, and SI would have a stronger effect on BI on mandatory
settings.Datasupportedtheseexpectedoutcomes.

Therestofthemoderatorshadcontraryeffectsontheoppositedirection.Strongereffect
appeared in women when they were expected in men, older when expected in younger
people, and early stages of experience when expected in people with more experience in
thetechnology(seeTable4.11).

Direct relationships cannot be interpreted in the presence of interactions. Therefore,

variablesbehaviorisneitherhypothesizednorevaluated.Table4.10showsasimpledashto
illustratethis.Thedirectionoftherelationshipininteractionsbetweendirectdeterminants
andmoderatorsisinterpretedaccordingtothecodificationofthemeasures.Forexample,a
negativerelationingenderrefersstrongereffectsonmen.Thisisbecausethecategorical
question was converted in a continuous scale where 1 represents men and 2 women
(Muthn 1984). Thus, a negative sign for the correlation coefficient of moderators in this
studycanbeinterpretedas:astrongereffectinmen,youngerpeople,lessexperiencedor
strongerinmorevoluntarysettings.

149

CRITERIAVALIDATIONBASEMODEL(WITHMODERATORS)
InternalHypothesesofUTAUT
ExpectedCriteria Outcome
Relationin
VariableX Stronger Stronger
theoryXBI Sig Dir Sig Dir
Effect Effect
PE Direct
EE Direct
SI Direct
FC DroppedFactor
GEN Moderator
AGE Moderator
EXP Moderator
VOL Moderator
GENxPE Moderator Yes () Men True False False
GENxEE Moderator Yes (+) Women False False False
GENxSI Moderator Yes (+) Women False False False
AGExPE Moderator Yes () Younger False False False
AGExEE Moderator Yes () Younger False True True
AGExSI Moderator Yes (+) Older False False False
AGExFC DroppedFactor
EXPxEE Moderator Yes () Earlystages False False False
EXPxSI Moderator Yes () Earlystages True True True
EXPxFC DroppedFactor
VOLxSI Moderator Yes (+) Mandatory False True True
Table4.11CriteriaValidationBaseModel(withModerators) (Source:Author)

4.12 TestHypothesis3

H3: The original model will have an acceptable fit with the data and will be
statisticallysignificantpartiallysupported.

Thebasemodel(seeFigure4.5)hadanacceptablemodelfit,whentestedwithmoderators
at < 5 (Wheaton et al. 1977), and without moderators at < 3 (Carmines &
McIver 1981). However, the probability of getting the level of fit achieved on the base
modelwith moderators and without moderatorsis less than 0.001. Therefore,
Hypothesis3ispartiallysupported.

150

When the test includes moderators, chi square ratio is =3.388, p=0.000, only
acceptable by the most relaxed standards ( < 5). In contrast, the same indicators
achieved acceptability =2.784, p=0.000 in the absence of moderators in a more
rigorousstandardoffit( <3).

Model fit probabilistic significance is evaluated with a p value for chisquare and RMSEA.
Chisquarepvaluegivesatestofexactfit(Browneetal.1993),undertheassumptionthat
themodelsarecorrect.Theprobabilityofgettingtheleveloffitachievedonthebasemodel
withmoderatorsandwithoutmoderatorsislessthan0.001(AMOSshow0.000).Notethatp
valuesforchisquareareadoppositetothecorrelationcoefficientspvalues.Verysmallp
values for chisquare meannot significantand they are not acceptable (Byrne 2010;
Tabachnick&Fidell2007).

Similarly, results for the root mean square error of approximation (RMSEA) presents an
improvementbyremovingthemoderationeffectsfromthebasemodel.TheRMSEAvalues
decreasefrom0.069to0.060,whichisacceptablebutnotoptimal.

ThepvalueforRMSEAisindicatedinAMOS21asPCLOSE,anditprovidesatestofclosefit.
PCLOSE is the probability of getting a value as large as the RMSEAs. Results for the base
modelwithmoderatorsshowaprobabilityforRMSEA<.001(AMOSshow.000)toachievea
RMSEAaslargeas.069.Inthemodelwithoutmoderators,theprobabilityincreasesto.171
for a RMSEA = .060. This shows a probabilistic improvement by removing moderators.
However,PCLOSEisnotacceptableineithercaseasitshouldbeavaluegreaterthan0.5.

Thisthesisusedtheratioofminimumdiscrepancydividedbydegreesoffreedom( )
andtherootmeansquareerrorofapproximation(RMSEA)totestitshypotheses.However,
otherindicatorsoffitareshowninFigure4.4andreportedindetailinTable4.13.Basedon
the criteria presented, the base model achieved acceptable fit, but unacceptable
probabilisticsignificance.Therefore,Hypothesis3ispartiallysupported.

This research used IBM SPSS AMOS 21.0.0 (Build 1178) to test the covariancebased
structural equation models. In the estimation of the discrepancy, the researcher selected

151

thefollowingoptions(Byrne2010):methodmaximumlikelihood,covariancessuppliedas
inputunbiased,covariancesselectedtobeanalyzedMaximumlikelihood,500random
permutations, bootstrap performance 1000 samples, 90 percentile confidence intervals,
90biascorrectedconfidenceintervals,andbootstrapML(MaximumLikelihood).

The base model in Figure 4.4 (upper side) shows the relationships in UTAUT, with and
without moderators. Firstly, the model without moderators is shown. Standardized
regression weights for performance expectancy (.20), effort expectancy (.24), and social
influence(.35)inrelationshipwithbehavioralintentionarestatisticallysignificantatpvalue
< 0.001. Also, the relationship between behavioral intention and actual behavior was
significant at such level, with a standardized regression weight .70. The squared multiple
correlation(equivalenttoR2),showedthismodelexplains42%ofthevarianceofbehavioral
intentionand48%ofthevarianceofactualbehavior.

Secondly,themodelwithmoderatorsisshowninFigure4.5(lowerside).Thepresenceof
moderators does not allow interpreting the direct relationships between performance
expectancy,effortexpectancy,socialinfluenceandbehavioralintention.Thisfigurerather
shows information about the moderators. Three, two and one stars indicate the level of
significance (p < .05, p < .01, and p < .001 respectively). The numbers between square
bracketsindicatethatthevalueinsidecannotbeinterpretedinpresenceofmoderators.The
numbersnearthetipofeacharrow,indicateastandardizedregressionweight.Thearrows
indicate hypothesized causal relationships; the direction of the arrow indicates which
variable causes the other. Arrows pointing to the middle of other arrow, indicate that
gender, age, experience and voluntariness are moderating the relationships described by
the lines they are pointing. Figure 4.5 shows that gender moderates the relationship
betweenperformanceexpectancyandbehavioralintentionwithastandardizedregression
weight of .064 which is significant at a p value < .05. In the case of the model with
moderators 51.8% of the variance of behavioral intention is explained and 45.5 of actual
behavior.ForinterpretationofthemoderatorsseeTable4.11.

152

 4.13 TestHypothesis4

H4: The extended models internal hypotheses will achieve criterionrelated

validitypartiallysupported.

TheobjectiveofthisthesiswastestingHypothesis4withcovariancebasedSEM.However,
high collinearity between habittechnology fit, performance expectancy and effort
expectancy could reduce the validity of the results (Farrell 2010; Fornell & Larcker 1981).
ThetestwithcovarianceSEMwasstillconducted,butthecorrelationcoefficientscouldnot
bevalidlyinterpreted.Inordertoprovidevalidresultsaboutcriterionrelatedvalidityother
analyseswereused.However,noneofthemprovidevalidslopemagnitudeordirectionin
thepresenceofdiscriminantvalidityconcerns.Therefore,notallthecriteriacouldbetested
fortheextendedstructuralmodelandHypothesis4wasdeclaredpartiallysupported(see
Figure4.6andTable4.12).

EXTENDEDMODEL:UTAUT&HTF

x df 3.388, p 0.000, GFI 0.949, CFI

0.971, RMSEA 0.069, PCLOSE 0.005. x df 3.601, p 0.000, GFI 0.708, CFI 0.814, RMSEA
Correlations:PEEE=0.560***, 0.072, PCLOSE 0.000.
PESI=0.523***,PEHTF=0.812***, Correlations:PEEE=0.560***,PESI=0.523***,PEHTF=
EESI=0.402***,EEHTF=0.753***, 0.812***,EESI=0.403***,EEHTF=0.756***,SIHTF
SIHTF=0.452*** =0.453***
[]indicatetheeffectcannotbeinterpretedduetopresenceofinteractionterm(moderator)
*pvalue<.05,**pvalue<.01,***pvalue<.001
Actual Behavior (AB), Behavioral Intention (BI), Effort Expectancy (EE), HabitTechnology Fit (HTF), Performance Expectancy
(PE),SocialInfluence(SI).
Modelspecificationbasedon(Venkateshetal.2003)
Figure4.6ExtendedModel:UTAUT&HTF (Source:Author)

153

Fornell and Larcker (1981) offers a valid alternative approach to structural equation
modeling in the presence of high collinearity, particularly in cases where discriminant
validity between two variables was not achieved, such as habittechnology fit and
performance expectancy. Following this alternatives significance of the relationships was
testedusingftests.

CRITERIAVALIDATIONEXTENDEDMODEL
Variable Relationintheory ExpectedCriteria Outcome
X XBI Sig Direction Sig Direction Support
HTF Direct Yes (+) True [?] Indefinite
PE Direct Yes (+) True [?] Indefinite
EE Direct Yes (+) True [?] Indefinite
SI Direct Yes (+) True [?] Indefinite
FTESTStatisticalSignificanceoftherelationships:

13
BI&AB,pvalue=1.7542610334x10
EE&BI,pvalue=5.6032186607x1006
HTF&BI,pvalue=1.6135821915x1005
PE&BI,pvalue=3.1618442630x1005
SI&BI,pvalue=7.2609026852x1003
Actual Behavior (AB), Behavioral Intention (BI), Effort Expectancy (EE), HabitTechnology Fit (HTF), Performance
Expectancy(PE),SocialInfluence(SI).
[?]Thedirectionoftheslopecannotbeinterpretedinthepresenceofmulticollinearity.
Table4.12CriteriaValidationExtendedModel (Source:Author)

Ftest evaluates the probability (two tail) of finding variance differences between each
variable and Behavioral Intention. The results presented in Table 4.12 indicate that the
variances of each pair of variables are not significantly different. This provides some
supporttheinternalhypotheses.Butitisnotenoughtodeterminethecompletebehavior
that theory would expect from these relationships. Therefore, Hypothesis 4 is partially
supported.

4.14 TestHypothesis5

H5: The extended model will have an acceptable fit with the data and will be
statisticallysignificantpartiallysupported.

Convergence and differentiation are not relevant to the properties of the chi square in
covariancebasedSEM.Fornell(Fornell&Larcker1981)demonstratesthatmodelfitcanbe

154

tested despite the assumption of multicollinearity. Thus, the test of Hypothesis 5 was
partially supported because the extended model achieved an acceptable fit, but its
particularspecificationfailedtoachievestatisticalsignificance.

The extended model with moderators achieved model fit at < 5 level. With slight
differences,thisappliedfortheextendedmodelwithmoderators( =3.388,p=0.000,
and RMSEA = 0.069, PCLOSE .005), and the extended model without moderators
( =3.601,p=0.000,andRMSEA=0.072,PCLOSE=0.000).Therewasanimprovementby
removing moderators from the model. However, the probabilistic values for model fit
remained<0.001(AMOSshowed0.000).Thereisanacceptablemodelfitwiththedatabut
the probability of a good fit in other samples of the same populations are very low.
Therefore,Hypothesis5ispartiallysupported(seeFigure4.6).Detailedmodelfitindicators
arereportedinTable4.16StructuralModelFitComparison.

The extended model reported similarities with the base model. The model fit showed a
significant improvement by removing the moderating variables. Furthermore, and
RMSEA resulted equivalent between the base model with moderators and the extended
modelwithoutmoderators( =3.388,p=0.000,andRMSEA=0.069).

This analysis also used IBM SPSS AMOS 21.0.0 (Build 1178) to test the covariancebased
structural equation models. In the estimation of the discrepancy, the researcher selected
thefollowingoptions(Byrne2010):methodmaximumlikelihood,covariancessuppliedas
inputunbiased,covariancesselectedtobeanalyzedMaximumlikelihood,500random
permutations, bootstrap performance 1000 samples, 90 percentile confidence intervals,
90biascorrectedconfidenceintervals,andbootstrapML(MaximumLikelihood).

4.15 Redundancyanalysis

Theredundancytestrevealedthatbykeepinghabittechnologyfitinthemodel,insteadof
performance expectancy and effort expectancy, the total loss on the effect size would be
0.4%.Thatlossistheeffectsizethatnoothervariableinthemodelcanprovideifthesetwo
155

variablesaredropped.Incontrast,bydroppinghabittechnologyfitauniquemarginof5.2%
oftheeffectsizeuponbehavioralintentionwouldbelost.

In the posthoc modification process, the effect size of each variable was calculated to
determinetheimpactofeliminatingtheoreticallyimportantvariablessuchasperformance
expectancyandeffortexpectancy.Aredundancytestwasdeemedforthispurpose(Fornell
&Larcker1981).PLSwasusedinthiscasetoavoidviolatingthemulticollinearityassumption
ofSEM(Byrne2010;Chin1995)

The coefficient of determination provides information on the percentage of the variance

explainedonadependentvariable.Thiscoefficientisfrequentlyreferredas inPLS,and
squaredmultiplecorrelation(SMC)incovariancebasedSEMwithAMOS.Table4.13showsa
comparisonofthecoefficientofdeterminationforthesixmodelstestedinthisstudy.

COEFFICIENTOFDETERMINATIONPLSVSSEM:BEHAVIORALINTENTION
BaseModel ExtendedModel ModifiedModel
BI SMC SMC SMC
Moderators .549 .5341 0.518 .596 .5792 .579 .578 .5667 .555
No
.378 .3742 .417 .427 .4123 .465 .423 .4206 .460
Moderators
=VarianceExplained(SmartPLS2.0)
1 1 =VarianceExplainedAdjusted
=samplesize
=regressionlines
SMC=SquaredMultipleCorrelation(IBMSPSSAMOS21.0.0)
andSMCpresentedinthisTablewerecalculatedforbehavioralintention(BI)
(SeemodelsspecifiedinSmartPLSinAppendix11,andUTAUTinPLSGraphinAppendix11)
PLS aims of increasing R2, thus greater values are considered to represent better models
(Hair,Ringle&Sarstedt2011).
Table4.13CoefficientofdeterminationPLSvsSEM:Behavioral (Source:Author)
Intention

PLSGraph 3.0, SmartPLS 2.0 and SPSS AMOS 21 were used to calculate coefficients of
determination.PLSGraph3.0andSmartPLS2.0didnotreportdifferencesin inthebase
modelwithmoderators.TherestofthevalueswerecalculatedonlywithSmartPLS2.0and
AMOS 21. Differences can be appreciated between R2 and SMC in Table 4.13. For the
redundancytestthevaluesfromPLSwerethenused.
156

REDUNDANCYANALYSIS:R2ANDEFFECTSIZEUPONBEHAVIORALINTENTION
Id Area
1 HTF 0.316 0.462
2 PE 0.23 0.299
3 EE 0.185 0.227
4 SI 0.269 0.368
5 HTFPE 0.335 0.504
6 HTFEE 0.323 0.477
7 HTFSI 0.423 0.733
8 PEEE 0.287 0.403
9 PESI 0.342 0.520
10 EESI 0.345 0.527
11 HTFPEEE 0.341 0.517
12 HTFPESI 0.425 0.739
13 HTFEESI 0.425 0.739
14 PEEESI 0.378 0.608
15 HTFPEEESI 0.427 0.745
A HTF 0.049 0.052
B PE 0.002 0.002
C EE 0.002 0.002
D SI 0.086 0.094
E HTFPE 0.031 0.032
F HTFEE 0.034 0.035
G HTFSI 0.005 0.005
H PEEE 0.000 0.000
I PESI 0.016 0.016
J EESI 0.004 0.004
K HTFPEEE 0.040 0.042
L HTFPESI 0.053 0.056
M HTFEESI 0.017 0.017
N PEEESI 0.001 0.001
O HTFPEEESI 0.087 0.095
=Union =EffectSize
=Intersection
DeterminationCoefficientforBehavioralIntention
Actual Behavior (AB), Behavioral Intention (BI), Effort Expectancy (EE), HabitTechnology Fit (HTF),
PerformanceExpectancy(PE),SocialInfluence(SI).
(BasedonByrne(2002)andPolkowski(2013),seeformulasinAppendix13)
Table4.14RedundancyAnalysis:R2andEffectSizeupon (Source:Author)
BehavioralIntention

157

Aredundancytestshowsthesharedportionsofthevarianceexplainedbytheindependent
variablesuponthedependentvariable.InFigure4.7,basedonByrne(2002)andPolkowski
(2013), a Venn diagram illustrates the contribution of each construct to behavioral
intentionscoefficientofdetermination.Thisiscalculatedonthebasisofsimpleunionsand
intercepts.Appendix4.6providestheformulasofthistest.

In the diagram, three circles represent performance expectancy, effort expectancy and
socialinfluence,andahoserepresentsthehabittechnologyfitconstruct(Figure4.7).Table
4.14 summarizes coefficients of determination and Effect Size values for
behavioralintention(BI).Areasidentifiedwithaletterrepresent(1)thecontributionofeach
variablewhichisnotsharedwithanyothervariableand(2)theinterceptsofthedifferent
variablesinthepossiblecombinations(Byrne2002;Rein1997).

Theresultsoftheredundancytestshowthatthehabittechnologyfitvariablecanreplace
performance expectancy and effort expectancy. Within this studys sample, this
replacementimpliesanegligiblelossonthevarianceexplained .Theunionofsections
B, H and C represent only 0.4 percentual points out of a total 42.7, i.e. less than 1%
of svalue.Incontrast,theIdArepresents4.5percentualpointsofthesametotal,i.e.

11.5%of svalue.Inconclusion,iftheresearcherhastodecidebetweenusingPE,EEor
HTF, these results indicate that the loss in variance explained would be greater is HTF if
dropped.Incontrast,thelossisnegligibleifPE,EEorbotharedroppedandHTFkept.

158

VENNDIAGRAMOFVARIANCEEXPLANATION

BasedonByrne(2002)andPolkowski(2013)
Figure4.7VennDiagramofVarianceExplanationR2 (Source:Author)

4.16 ModifiedModel

ThemodelextendedwasmodifiedbecauseResearchQuestion3relatestothepossibilityof
improvingthemodelspecificationofUTAUT.Thisquestionwasformulatedfromtheearly
stages of this research. However, after conducting analysis of the extended model it was
foundproblematic.Ifhabittechnologyfitmayplayaroleinit,themodelmustbemodified.
Inthiscaseitisimperativetoselectwhichconstructsshouldbeexcluded,becauseitisnot
conceptuallysensibletocombinethesefactors(Farrell2010).

159

Intheinitialstrategy,therewasamodificationplanningstageaftergettingtheresults.This
processincludedgoingbacktotheliteraturetoguidethemodificationaccordingtotheory.
Theunexpecteduseofaredundancytestprovidednewinformationabouttherelativevalue
of habittechnology fit against performance expectancy and effort expectancy. A Venn
diagram showed that habittechnology fit contained the effect size of the other two
constructs almost completely, plus an additional margin. Dropping habittechnology fit
wouldmeanlosing5.2%oftheeffectsize,whereasexcludingperformanceexpectancyand
effortexpectancytogetherwouldsignifyonlyalossof0.4%.Thisresultdoesnotdenythe
importance of any of the constructs; it just reveals they are redundant, and a new model
canbeproposedwithoutlosingexplanatorypower(Byrne2002;Polkowski2013).

CFAwasconductedonthefactorsofthenewlyvisualizedmodel,andnoconcernsemerged.
Then, the new model was specified in SEM (and secondarily in PLS). The new model
achievedoptimalmodelfitandchisquaregotstatisticalsignificance.

Every model has been tested with and without moderators. The reason is that it is not
possible to give interpretation to the direct relationships in the presence of moderators.
Therefore, the evaluation has to be done in two steps, with and without moderators
(Whisman & McClelland 2005). At the end, six models in total were evaluated (base,
extended,andmodifiedwithandwithoutmoderatorsineverycase).Figure4.8showsthe
specificationofthemodelthatachievedthebestmodelfit.ThemodifiedmodelorHabit
Technology Fit Model (HTF Model) achieved optimal fit at < 2 level, as well as
statisticalsignificance( =1.545, = 0.100,RMSEA=0.033, PCLOSE=0.822).These
values are acceptable at the most rigorous level, and are appropriate to confirm theory
(Arbuckle2010;Byrne2010).

This analysis was conducted with IBM SPSS AMOS 21.0.0 (Build 1178) with the following
options: method maximum likelihood, covariances supplied as input unbiased,
covariances selected to be analyzed Maximum likelihood, 500 random permutations,
bootstrap performance 1000 samples, 90 percentile confidence intervals, 90 bias
correctedconfidenceintervals,andbootstrapML(MaximumLikelihood)(Byrne2010).

160

 NEWSPECIFICATION:HABITTECHNOLOGYFITMODEL

x df 1.545, p 0.100, GFI 0.990, CFI 0.996, RMSEA 0.033, PCLOSE 0.822.
Correlation:SIHTF=0.448***
*pvalue<.05,**pvalue<.01,***pvalue<.001
ActualBehavior(AB),BehavioralIntention(BI),HabitTechnologyFit(HTF),SocialInfluence(SI).
Figure4.8NewSpecification:HabitTechnologyFitModel (Source:Author)

Figure4.8showsthespecificationofthestructuralmodelfortheposthocmodifiedmodel.
Therelationshipbetweenhabittechnologyfitandbehavioralintentionispositive( =.44)
andsignificantatpvalue<0.001.Therelationshipbetweensocialinfluenceandbehavioral
intentionresultedpositive( =.36)andsignificant(pvalue<0.001).Finally,therelationship
betweenbehavioralintentionandactualbehaviorwasalsopositive( =.70)andsignificant
atpvalue<0.001.Habittechnologyfitandsocialinfluenceexplained46%ofthevarianceof
behavioralintention,andbehavioralintention48%ofactualbehavior.

4.17 TheoreticalCriteriaforPosthocModification

The proposed HTF Model is a result of the posthoc modification of UTAUT and UTAUT
extendbyhabittechnologyfit.Table4.12showsevaluationsoftheorybasedcriteria.This
assessmentprovidedsupporttohabittechnologyfitandsocialinfluenceinrelationshipwith
behavioralintention.Ageandexperiencemoderatorrelationshipsweresupportedforhabit
technology fit, and experience for social influence. The relationship between behavioral
intentionandactualbehaviorwassupported.Thestructuralmodelforthistestisshownin
Figure4.6.

161

 CRITERIAVALIDATIONMODIFIEDMODEL
(WITHMODERATORS)
ExpectedCriteria Outcome
Relationintheory
Variable Stronger Stronger
XBI Sig Dir Sig Dir Supported
Effect Effect
HTF Direct Yes (+) N/A True True N/A Yes
SI Direct Yes (+) N/A True True N/A Yes
GENxHTF Moderator Yes () Men False True True No
GENxSI Moderator Yes (+) Women False False False No
AGExHTF Moderator Yes (+) Older True True True Yes
AGExSI Moderator Yes (+) Older False False False No
EXPxHTF Moderator Yes (+) Laterstages True True True Yes
EXPxSI Moderator Yes () Earlystages True True True Yes
VOLxSI Moderator Yes (+) Mandatory False True True No
Relationintheory Stronger Stronger
Variable Sig Dir Sig Dir Supported
XAB Effect Effect
BI Direct Yes (+) N/A True True N/A Yes
ActualBehavior(AB),Age(AGE),BehavioralIntention(BI),EffortExpectancy(EE),Experience(EXP),Gender(GEN),Habit
TechnologyFit(HTF),PerformanceExpectancy(PE),SocialInfluence(SI),Voluntariness(VOL).
Table4.15CriteriaValidationModifiedModel(withModerators) (Source:Author)


POSTHOCMODELMODIFICATIONINPROCESS

x df 3.507, p 0.000, GFI 0.822, CFI 0.868, RMSEA 0.071, PCLOSE 0.000.
Correlation:SIHTF=0.448***
[]indicatetheeffectcannotbeinterpretedduetopresenceofinteractionterm(moderator)
*pvalue<.05,**pvalue<.01,***pvalue<.001
ActualBehavior(AB),Age(AGE),BehavioralIntention(BI),Experience(EXP),Gender(GEN),HabitTechnologyFit(HTF),
SocialInfluence(SI),Voluntariness(VOL).
Figure4.9PosthocModelModificationinProcess (Source:Author)

162

ThemodelintheprocessofmodificationinFigure4.9wasusedtoassessthemoderators.
The direct relationships standardized coefficient weights cannot be interpreted in the
presenceofmoderators(forthatpurposeseeFigure4.8).Resultsshowthatgenderisnot
significant in its moderating relationship with habittechnology fit and social influence to
behavioralintention.Ageresultedstatisticallysignificantatpvalue<0.05asamoderatorof
habittechnologyfitandsocialinfluencetobehavioralintention.However,thestandardized
coefficientweightoftheinteraction(ageandsocialinfluence)issmallandnegative( =
0.07), indicating slightly stronger effect of social influence upon behavioral intention for
younger. The coefficient for the interaction of age and habittechnology fit is positive and
greaterinmagnitude( =0.17),stillconsideredasmalleffect.Experienceasamoderator
was significant at p value <0.001 for habittechnology fit and social influence upon
behavioral intention. The stronger moderation effect of experience was upon social
influence and behavioral intention ( = 0.17) and smaller for habittechnology fit upon
behavioral intention ( = 0.11). This is interpreted as followsthe relationship between
socialinfluenceandbehavioralintentionisstrongerforindividualswithlessexperience,in
the first case. In the second case, the relationship of habittechnology fit upon behavioral
intention is stronger for more experienced individuals. Voluntariness, as a moderator,
resultednotstatisticallysignificant.

4.18 ComparativeModelFit

Theprocessofposthocmodelmodificationincludedmodelcomparison.Strictlyguidedby
thetheoreticalrelationshipssuggestedbyUTAUTandtheonebetweenhabitandintention,
severalmodelsweretested.However,Table4.16reportsthemostimportantcombinations.
The best fit was achieved by the modified model ( = 1.545), followed by UTAUT
withoutmoderators( =2.784),extendedmodelwithoutmoderatorsandUTAUTwith
moderators ( = 3.338), modified posthoc model with moderators ( = 3.507),
extendedmodelwithmoderators( =3.601).Thus,theposthocmodelisproposedas
thebestunderthiscriterion.

Thesamesoftwareandanalysisoptionswereusedtotestthecovariancebasedstructural
equation models: IBM SPSS AMOS 21.0.0 (Build 1178), method maximum likelihood,
163

 covariancessuppliedasinputunbiased,covariancesselectedtobeanalyzedMaximum
likelihood,500randompermutations,bootstrapperformance1000samples,90percentile
confidence intervals, 90 biascorrected confidence intervals, and bootstrap ML (Maximum
Likelihood)(Byrne2010).

STRUCTURALMODELFITCOMPARISON

UTAUTBaseModel UTAUTExtended ModifiedModel Benchmark

Amodelhasa
Without Extended Without Modified Without
INDICATOR BaseModel good fit if
moderators Model Moderators Model Moderators
(Byrne2010):
(BI)SMC 0.518 0.417 0.579 0.465 0.555 0.460

3154.361 61.246 6082.21 152.45 1932.43 18.546

931 22 1689 45 551 12

3.388 2.784 3.601 3.388 3.507 1.545 2

0.000 0.000 0.000 0.000 0.000 0.100 0.05

SRMR 0.134 0.0316 0.1598 0.0361 0.113 0.024 SRMR 0.05

GFI 0.782 0.974 0.708 0.949 0.822 0.99 GFI 0.90

AGFI 0.758 0.946 0.684 0.912 0.797 0.977 AGFI 0.80

PGFI=0poorfit;
PGFI 0.703 0.476 0.654 0.547 0.719 0.424 PGFI=1goodfit
NFI 0.831 0.976 0.76 0.959 0.825 0.99 NFI 0.90

CFI 0.874 0.984 0.814 0.971 0.868 0.996 CFI 0.90

RMSEA 0.05
good fit; 0.05
RMSEA 0.08
acceptablefit;
RMSEA 0.069 0.06 0.072 0.069 0.071 0.033 0.08 RMSEA
0.10 marginal
fit;RMSEA
0.10 poor fit.
PCLOSE 0.000 0.171 0.000 0.005 0.000 0.822 PCLOSE 0.50

HOELTER HOELTER (.05)

160 279 148 204 158 570 200
.05
HOELTER HOELTER (.01)
165 331 151 231 164 710 200
.01
Table4.16StructuralModelFitComparison (Source:Author)

164

4.19 BriefReportofProblemsinAnalysis

Using the method for interaction shown in Figure 4.9.1 (Iacobucci 2010), the results
between the coefficient of determination ( ) and Squared Multiple Correlation (SMC)
showed small discrepancies. In contrast, by using the procedure to specify Whisman &
McClellands(2005)proceduresforinteraction,discrepanciesweresignificantlygreater.

When the interactions were calculated by direct multiplication (Whisman & McClelland
2005)incovariancebasedSEM,AMOS21.0producesaveryhighSMCvalue,butaverybad
fit of the model. That was not a problem in PLS where direct multiplication and mean
deviationmultiplied(Iacobucci2010)producedthesameresult.

Withdirectmultiplication,thebasemodelwithmoderatorsobtainedaSMC=.714,similarto
Venkateshs (2003) R2 values. However, the fit of the model ( = 57.083) was most
unacceptable from any standard (Byrne 2010; Carmines & McIver 1981; Wheaton et al.
1977).ByspecifyingmoderationeffectsbydirectmultiplicationinAMOS,thereisariskof
typeIIerror.

4.20 Summary

Thethreephasesoftheanalysiswere(1)datapreparation,(2)reliabilitytest,exploratory
factor analysis (EFA) and confirmatory factor analysis (CFA), and (3) validity assessment,
model testing and model modification. This section presented the outcome of the
hypotheses of this study. Six models were tested: base, extended and modified (with and
withoutmoderators).

Therelationshipofhabittechnologyfitandbehavioralintentionwastested.Inallthecases
the relationship results were positive and significant (p<.001). This results support
hypothesis 1. Results supported age and experience as moderators, but not gender.
Therefore, Hypothesis 1a was partially supported. The base model achieved criterion
relatedvaliditywithoutmoderator,butnotwiththem.Italsoachievedacceptablemodelfit
whentestedwithmoderatorsat <5,andwithoutmoderatorsat <3,butlow
probabilityofgoodfitinothersamples<0.001.
165

Posthocmodelmodificationwasconductedbasedonthebaseandextendedmodels.The
modifiedmodelorHabitTechnologyFitModel(HTFModel)achievedoptimalfitat <
2level,aswellasstatisticalsignificance( =1.545, =0.100,RMSEA=0.033,PCLOSE
= 0.822). This model fit is acceptable at the most rigorous level, and is appropriate to
confirmtheory.

166

 CHAPTER5 FINDINGSANDDISCUSSION
5.1 Objective

Thepurposeofthischapteristodiscussthekeyempiricalandsomeadditionalfindingsof
thisresearch.Thischapterexaminestheresultsgeneratedbythedataanalysis,andrelates
themtopreviousresearchofhabitandtechnologyacceptance.Thischapteralsopresents
theoreticalcontributionsofthisresearch.

5.2 Positiverelationshipofhabittechnologyfitandbehavioralintention

The results of this thesis supported a positive and statistically significant relationship
between habittechnology fit and behavioral intention. This finding confirms previous
technology acceptance studies where a positive relationship between single habit and
behavioral intention was supported, such as the study on data mining tools presented in
Huang,WuandChou(2013),theresearchaboutmobileInternetinVenkatesh,Thongand
Xu (2012), virtual worlds in Barnes (2011), ecommerce adoption in Liao, Palvia and Lin
(2006),andbusinesstoconsumerwebsitesinGefen(2003)suggestedtherelationshipofa
single habit and behavioral intention. This thesis extended the measurement capacity in
these studiesfrom single to multiple and from predetermined to nonpredetermined
habits. This thesis makes a novel contribution to the literature of habit, technology
adoption, personenvironment fit and compatibility by offering the conceptualization of
HabitTechnology Fit as a new construct, and the systematical development of the
correspondingmeasurementscale.

Therelationshipofhabitandbehavioralintentionshasrarelybeenstudiedincludingother
thansinglepredeterminedhabitthatcorrespondstoitstargetbehavior.Thisthesisextends
previous research by providing a new measurement that considers multiple non
predetermined habits, which constitutes an original and significant contribution to the
literatureofhabitandtechnologyacceptance.

Byextendingthescopeofthehabitsthatcanbecapturedwiththenewmeasurementscale,
thisthesisalsoextendsprevioustheoryofhabitsbyaddressingthegapabouttheinfluence

167

of a structure of multiple habitsnot just oneupon behavioral intention. Therefore, an

originalandsignificantcontributionofthisthesistotheliteratureofhabitandtechnology
acceptance is the incorporation of multiple nonpredetermined habits to a technology
acceptancemodel.ThefindingsofthisthesisconfirmstudiessuchasChenandLai(2011)on
public transport use, Klckner, Matthies and & Hunecke (2003) on travel mode choices
(usingbipolarmeasureandmultipleRFM),VerplankenandFaes(1999)onunhealthyfood
habits, Verplanken (1994) on car use, and Landis, Triandis & Adamopoulos (1978) on
classroom teacher behavior. Although the habits considered in these studies were
predeterminedbytheirresearchers,theirmeasurescapturedmultiplehabits.Thesestudies
hypothesizedthepositiverelationshipbetweenhabitandbehavior,butimplicitlytestedthe
relationshipofmultiple,stillpredetermined,habitsandbehavior.

The measurement technique found in Verplankens Study 4 (Verplanken & Orbell 2003)
asked participants to create a list of habits and report their frequency, and then the
researcherwouldtakethehabitwithhighestfrequency.TheworkofStudy4wouldendina
finalmeasureofasinglehabit,butithadthevalueofnotbeingimposedbytheresearcher.
Thus, this thesis extends the scope of nonpredetermined habit measurement by
incorporatingameasureformultiplehabits.

This research confirm the positive relationship of compatibility upon behavioral intention
previously observed in personenvironment fit studies (Cable, D. M. & DeRue, D. S. 2002;
KristofBrown, Zimmerman & Johnson 2005), and compatibility dimensions upon behavior
in technology acceptance (Karahanna, Agarwal & Angst 2006). It may also extend
Karahanna, Agarwal and Angsts (2006) acceptance model as habit technology fit could
provideanewdimensionofcompatibility.

Validationoftheoreticalconceptsthathadnotbeentestedempiricallybeforeisprovidedin
this thesis. It confirms Bourdieu (1984) and Hodgsons (2010) theoretical studies in that
Individualsrelyontheirhabitusasatooltorespondtochanges(Bourdieu1990,p.290),
and in that rational choices themselves are always necessarily reliant on prior habits
(Hodgson2010,p.6).

168

5.3 Moderation of age, experience and gender upon habit habittechnology fit and
behavioralintention

The relationship between habittechnology fit and behavioral intention was significantly
moderatedbyageandexperiencebutnotbygender.Theeffectofhabittechnologyfitupon
behavioralintentionwasstrongerintheolderandmoreexperienced.Genderwasslightly
strongerformen,butstatisticallynotsignificant.

These findings would confirm age and experience act as significant moderators in the
relationshipbetweenhabitandbehavioralintentioninpreviousempiricalresearch,suchas
in the study on mobile Internet technology (Venkatesh, Thong & Xu 2012), and the
importanceofageinelectronicbankingacceptance(Dabholkar&Bagozzi2002).Theyalso
confirm similar empirical results in the study of acceptance of 3Dgestures (Comtet 2013)
basedinwhichgenderappearsnotasignificantmoderatorofacceptance.

In regards to age and habit, the findings of the main study may also empirically provide
somesupporttotheideasofDewey(2002)whosuggestedthatadultcustomandhabitsget
strongerwhilegrowing,arejealouslykept,andtametheoriginalityoftheyoungerself.In
otherwords,italsosupportthenotionsabouttheconservativepullofhabitsinWoodand
Neal(2009),andthatoldhabitsarelikelytobemaintainedbecauseofthespeedandease
with which past patterns of behavior can be initiated and executed (Ouellette & Wood
1998).

Forgender,Venkatesh,ThongandXu(2012)carefullyarticulatedanargumentinorderto
hypothesize its relationship with habit. The argument included strong propositions from
psychologyandconsumerresearchsuchas(Gilligan1982),(Farina&Miller1982),(Krugman
1966),and(MeyersLevy&Tybout1989)alsofoundin(Goldner&Levi2014),(Iijimaetal.
2001),and(Milne&Greenway1999).Thesepropositionssuggestthatgenerallywomenwill
showhigherlevelsofattentiontodetailcomparedtomen.Intheirargument,Venkatesh,
Thong and Xu (2012) suggest that the greater the attention to detail, the smaller the
attachmentfortheownhabits.Therefore,ifwomenpaymoreattentiontodetailthanmen,
theyshouldbelessattachedtotheirhabits.Thefindingsofthisthesisinregardstogender

169

as a moderator of habit may empirically reject this argument and may confirm the
qualitativeresearchinAlHtaybatandvonAlbertiAlhtaybat(2013)abouttheirrelevantrole
ofgenderintechnologyadoption.

5.4 Basemodel

The results of this thesis showed that not all the original relationships in Venkatesh et al.
(2003)obeythetheoreticalcriteriasuggestedbytheUnifiedTheoryofAcceptanceandUse
of Technology (UTAUT). The key independent variables of the modelperformance
expectancy, effort expectancy, and social influenceheld a positive and significant
relationshipwithbehavioralintention.Therewasalsoapositiveandsignificantrelationship
betweenbehavioralintentionandactualbehavior(facilitatingconditionswasdroppedafter
thefactoranalysis,andcouldnotbeincludedinthetest).However,themoderatorsdidnot
achievecriterionrelatedvalidity,leadingtosupportHypothesispartially.

5.4.1 KeydeterminantsofintentioninUTAUT

These findings confirmed a positive relationship between three key determinants and
behavioral intention: performance expectancy, effort expectancy, and social influence,
which previous research had found such as (Venkatesh et al. 2003; Venkatesh & Zhang
2010)inUTAUTandUTAUT2wherethefullunmodifiedmodelwastested,and(Venkatesh,
Thong & Xu 2012) where an extended model is tested. Similar findings were present in
numerousstudieswhereallthekeydeterminantsareallsignificant.Examplescanbefound
inNeys(2013)thesis onmCRMapplicationsforsmartphones,inthe Powelletal.(2012)
study on largescale enterpriselevel systems, in Sok Foon and Chan Yin Fahs (2011)
researchonInternetbankingtargetingindividualsoutofanorganization,theworkofWang
et al. (2010) on the acceptance of distance learning technologies, McLeod, Pippin and
Catanias(2009)studyonTaxSoftwareuse,andBandyopadhyayandFraccastoros(2007)
researchonprepaymentmeteringsystems.

These findings also confirm research prior to UTAUT (Venkatesh et al. 2003), where the
constructs might be considered quite equivalent as they share some or all the same

170

measurement items. The constructs in which performance expectancy has its roots are
perceived usefulness (Davis 1989) from where performance expectancy inherited all four
measurement items. Other constructs associated to performance expectancy are extrinsic
motivation (Davis, Bagozzi & Warshaw 1992), Jobfit (Thompson, Higgins & Howell 1991),
Relative Advantage (Moore & Benbasat 1991), and Outcome Expectations (Compeau &
Higgins 1995). These constructs are related to the expectations of the users who believe
thatusingtechnologywillenhancetheirperformance,helpthemtoachieve,offertheman
advantage or improvement to their outcomes. Thus, this thesis confirms performance
expectancy and its predecessors positive and significant relationship with behavioral
intention.

Thisthesisalsoconfirmsthepositiveandsignificantrelationshipbetweeneffortexpectancy
and behavioral intention (Venkatesh et al. 2003). Effort expectancy is equivalent to
Perceived Ease of Use (Davis 1989; Moore & Benbasat 1991) from which its measures
derive.AnothersimilarconstructisComplexity(Thompson,Higgins&Howell1991)because
all these constructs reflect the individuals estimation about ease or difficulty that will be
involvedinusingatechnology.

Social influence (SI) (Venkatesh et al. 2003) has its roots in a great tradition of research.
Thus, the results of this thesis confirm the positive and significant relationship between
social influence and behavioral intention. Social influence derives from: Subjective Norm
(Ajzen 1991) from where two UTAUT items derive, Social Factors (Thompson, Higgins &
Howell1991)whichprovidesUTAUTwiththeothertwomeasurementitemsthatintegrate
the social influence scale. The influence of others through their opinions and status in a
social system are contained in these constructs, and they considered important
determinantsinthedecisionofusingtechnology.

5.4.2 ModeratorsoftheUTAUTbasemodel

Thefindingsofthisthesisrejectedage,genderandvoluntarinessasmoderatorsofUTAUT.
Only experience was confirmed to moderate the relationship of social influence and
behavioralintentionwithstrongereffectwithlessexperiencedindividuals.

171

Forageasamoderator,thisthesisdiffersfromthefindingsofVenkateshetal.(2003)and
VenkateshandZhang(2010)intheUSandBandyopadhyayandFraccastoro(2007)inIndia,
butitsupportspreviousresearchfromMartins,OliveiraandPopovi(2014)inPortugal,Chu
(2013) in China and Taiwan, Ney (2013) in the US, Venkatesh and Zhang (2010) in China,
AbuShanab and Pearson (2007) in Jordan, AlGahtani, Hubona and Wang (2007) in Saudi
Arabia,aswellasLu,YuandLiu(2009)inChina.

For gender as a moderator, this thesis refutes Venkatesh et al. (2003), Venkatesh and
Zhangs(2010)findingsintheUSandBandyopadhyayandFraccastoro(2007)inIndia,but
confirms the findings in the research of Martins, Oliveira and Popovi (2014) in Portugal,
Chu (2013) in China and Taiwan, Ney (2013) in the US, Powell et al. (2012) in the US,
VenkateshandZhang(2010)inChina,Wangetal.(2010)inTaiwan,Lu,YuandLiu(2009)in
China,AlGahtani,HubonaandWang(2007)inSaudiArabia,andLin,ChanandJin(2004)in
Singapore.

Forvoluntarinessasamoderator,thisthesisdivergesfromthefindingsofVenkateshetal.
(2003) and Venkatesh and Zhang (2010) in the US and Bandyopadhyay and Fraccastoro
(2007) in India, but this thesis confirms previous results from Chu (2013) in China and
Taiwan,VenkateshandZhang(2010)inChina,SahuaswellasGupta(2007)inKoreaandthe
US.

Whereas UTAUT hypothesized significance and a particular direction of the moderation

effectsofage,genderandvoluntariness(Venkateshetal.2003),thisthesisfindingsshow
either lack of significance, a different direction (stronger effect in males when it was
expectedinfemales)orboth.VenkateshandZhang(2010)suggestedthatthedivergences
from the original UTAUT in the cases of failure for gender, age and voluntariness can be
attributed to cultural differences (Hofstede 1983)specifically to individualism and
collectivism.Suchpropositionwouldbesustainedbytheresearchconductedincountries
identified as collectivists like China (Lu, Yu & Liu 2009; Venkatesh & Zhang 2010), Jordan
(AbuShanab&Pearson2007),SaudiArabia(AlGahtani,Hubona&Wang2007),Singapore
(Lin,Chan&Jin2004),andTaiwan(Wangetal.2010).However,iftheoriginalversionof
UTAUTismoresuitableforindividualistculturesasVenkateshandZhang(2010)affirm,such
172

apropositionwouldnotexplainthefailureofthemoderatorsintheUS(Powelletal.2012)
ortheirsuccessinIndia(Bandyopadhyay&Fraccastoro2007).

The findings of the present study, in contrast to previous research conducted in different
culturalsettings,haveshownmixedresultsoninteractionsofgender,ageandvoluntariness.
These inconsistencies make the outcomes of Hypothesis 2 of extreme relevance, as they
suggest extremely low probability that UTAUT structural model with moderators will find
goodfitalongothersamplesofthesamepopulation.

5.4.3 Basemodelfitwithdata

The results of this thesis indicated acceptablemodel fit for the base model, yetsuch a fit
wasfoundveryunlikelytofindthesameorbettermodelfitinothersamplesofthesame
population.

Soundtheoryisexpectedtoconsistentlybeagoodmatchbetweenthestructuralspecified
relationshipsanddatafromempiricalobservations.CovariancebasedSEMisappropriateto
confirm theoretical models, testing unique structures of theoretical relationships as a
whole.TwoindicatorsincovariancebasedSEMprovideprobabilisticinformationaboutthe
fitofthemodelChisquareratiospvalueandRMSEAsPCLOSE(Byrne2010;Marsh,Balla
&McDonald1988).Theseindicatorsarefrequentlyoverlooked,andsometimesconsidered
unrealistic to achieve. But, they are as important as the p value that indicates the
significanceofacorrelationweightwhenitcomestoevaluatecompletemodels.

Anoriginalcontributionofthisthesistotheliteratureoftechnologyacceptancemayderive
from the analysis of the unmodified UTAUT base model using covariancebased SEM and
reportingprobabilisticvaluesforthemodelfit.UTAUThastraditionallybeenanalyzedwith
variancebasedStructuralEquationModelling(PartialLeastSquares),whichismoresuitable
forpredictionandexploratoryanalysis(Hair,Ringle&Sarstedt2011).Thecovariancebased
SEM comes with intricacies that makes proper analysis much more problematical than it
wouldbeforPLS.Thisisduetoitsassumptionsandtherelativelycomplexapproachtothe

173

moderating effects. Therefore, it is not surprising to find less studies analyzed with
covariancebasedSEMforUTAUT.

A study on distance learning technologies (Wang et al. 2010) included the UTAUTs key
independentvariablesforbehavioralintention,anditincludedgenderasamoderator.The
model achieved a Chi square ratio = 2.92, and RMSEA= .073. A study on mobile
technology (Wang & Wang 2010) used the original measurement scales, but added other
three independent variables upon behavioral intention, it achieved a Chi square ratio
= 1.43, and RMSEA = .04. Other study on prepayment metering systems
(Bandyopadhyay&Fraccastoro2007)specifiedamodelidenticaltoUTAUT,butitincluded
incomeasmoderator.ThismodelreportedaChisquareratio =3.96,andRMSEA=
.073.However,itisnotoriousthatnoneofthesestudiesprovidedprobabilisticvalues.

Thestudiesinthepreviousparagraphreportsimilarresultstotheonesofthisthesisinthat
they include UTAUTs variables in their studied models, Bandyopadhyay and Fraccastoro
(2007)reportthemostsimilarspecificationtoUTAUTbasemodelwithmoderatorsastested
in this thesis, and report very similar results (this thesis: = 3.338; Bandyopadhyay
andFraccastoro(2007): =3.96).

ThisthesisfoundbetterfitforUTAUTwithoutmoderators.ThisfindingsshowedthatUTAUT
without moderators achieved fit and probabilistic significance only for RMSEA ( =
2.784, p = 0.000, and RMSEA= .06, PCLOSE=0.171). The findings on the specification of
UTAUT without moderators are somehow similar, in complexity and variables, to more
parsimoniouspredecessorssuchasTRA(Ajzen&Fishbein1980),TAM(Davis1989)andthe
integrationofeaseofuse,usefulnessandsubjectivenorm(analyzedwithcovarianceSEM)
in(Suksangiam&Chaiyasoonthorn2013).

While previous research has used covariance based SEM in their analysis (see
Bandyopadhyay & Fraccastoro 2007; Wang et al. 2010; Wang & Wang 2010), and they
provide some evidence of the validity of UTAUT, their results are not fully comparable
becausetheyincludeotherfactorsinthestructuralspecification.Comparisonwouldrequire
studies testing UTAUT before extending or modifying it. Partial and modified versions of

174

UTAUTwouldnotallowaccuratecomparisons,becausewhenusingcovariancebasedSEMa
singlevariablecanalterthemodelfitradically.

ThisthesisagreeswithVenkateshandZhang(2010)inregardstothefullmodelrarelybeing
replicated faithfully (except in Venkateshs work). Besides, it was not possible to find
analysis reports of UTAUT being tested with covariancebased SEM in its strict original
specification.

Thisthesisismakingasignificantcontributiontotheliteratureoftechnologyacceptanceby
testing UTAUT faithfully. Although, facilitating conditions factor was not included due to
convergent and discriminant validity and its moderators were not tested in consequence.
Thebasemodel,correspondingtothefirstequationofUTAUTwastestedunmodified(See
Figure2.4UTAUTModel).

Thisthesisusedexactspecificationofthebasestructuralmodel,samescales,assecondary
techniquePLSwasused(originalanalysistechnique)andslightlynewerversionofsoftware
wasused(PLSGraph3.0).Besidesitwastriangulatedwiththemainanalysistechniqueof
this thesis (covariancebased SEM). Thus, the method followed makes the findings
appropriateforcomparisons.

Althoughlimitedtothebasemodel(
),

replicating the exact specification of UTAUT was an important task. Testing UTAUT base
modelbyaconfirmatoryanalysismethodsuchascovariancebasedSEMmaycontribute
withausefulbenchmarktotheliteratureoftechnologyacceptance.

5.5 Extendedmodel

The inclusion of habittechnology fit in the base model revealed a highly collinear
relationshipbetweenhabittechnologyfit,performanceexpectancyandeffortexpectancy.
Very similar findings were found in previous research. Discrimination issues have been
reported between performance expectancy (perceived usefulness) and individuals
compatibilitywithworkstylescales(whichisadimensionofcompatibility,andmightalso
175

be a dimension of habit). As in this thesis, which provided strong evidence of semantic

differentiation between concepts habittechnology fit, performance expectancy and effort
expectancy;Karahanna,AgarwalandAngst(2006)foundsignificantconceptualdifferences
between compatibility workstyle and perceived usefulness (same measurement as
performance expectancy). Still, factor analysis presented them as undifferentiated. In the
case of habittechnology fit, further analysis revealed an explanatory power of intention
greater than performance expectancy and effort expectancy combined. This finding may
contribute to revisit previous measurement of compatibility which has faced high
collinearitywithperformanceexpectancyandmaydeserveasecondevaluationsuchasin
workstylecompatibilityin(Karahanna,Agarwal&Angst2006).

Anexplanationforsuchparadoxicalfindingonthevalueofhabittechnologyfitreliesinthe
theory of measurement. Fornell and Larcker (1981) explain that convergent validity is
grantedwhenwithinconstructcorrelationsarehighandaboutthesamemagnitude.Onthe
otherhand,discriminantvalidityisconferredwhenthecrosscorrelationsarehigh,uniform,
andlowerthanthewithinconstructcorrelations.However,thecontentvalidityofthescale
cannotbedirectlyinferredfrompurestatisticmethodssuchasreliabilitiesorfactoranalysis
(DeVellis 2012). Thus, this finding may also constitute a significant contribution to the
measurement of habitrelated compatibilities in the literature of habit, compatibility and
technologyacceptance.

5.5.1 Extendedmodelfitwithdata

The specification of the extended model was found to have a slightly worse fit with data
than the base model ( = 3.601 vs = 3.388 respectively). The extended model
achieved an acceptable fit, but its particular specification failed to achieve statistical
significance as a whole, exactly as it happened with the base model (p = 0.000 and
PCLOSE=0.000 in both cases). Convergence and differentiation are not relevant to the
propertiesofthechisquare(Fornell&Larcker1981),andthereforeavalidresultonthefit
ofthemodelwasobtained.

176

The findings on the structure of the extended model may support model specifications of
previous research. Similar specifications where single habit was integrated habit with
UTAUTcanbefoundinEscobarRodrguezandCarvajalTrujillo(2013),Ney(2013),Pahnila,
SiponenandZheng(2011),Venkatesh,Thong andXu(2012).Thisthesissupports tosome
extentthemodelspresentedintheirwork.However,onlytheexactspecificationscouldbe
comparedorconfirmedwhendealingwithwholestructures,andnostudywasfoundwith
the exact same structure. The findings on the specific structural relationships of UTAUT
extendedbytheconstructdevelopedinthisthesis(habittechnologyfit)extendtheoryand
constituteanoriginalcontributiontotheliteratureofhabitandtechnologyacceptance.

Habittechnologyfit(HTF)wasnewconstructdeveloped,measuredandtestedinthisthesis.
Datageneratedfrom251adultsin25countriesconfirmedthefundamentalrelationshipsof
UTAUT and its integration with HTF in an extended model proposed in this thesis. The
quantitativedatastrengthensthisinvestigationandconfirmsthathabittechnologyfithasa
positive impact upon behavioral intention. This constitutes a novel contribution to theory
andtotheliteratureonhabit,andtechnologyacceptance.

5.6 Modifiedmodelandotherfindings

Thisstudyfoundthatbykeepinghabittechnologyfitinthemodel,insteadofperformance
expectancyandeffortexpectancy,thetotallossontheeffectsizewouldbe0.4%.Thatloss
istheeffectsizethatnoothervariableinthemodelcanprovideifthesetwovariablesare
dropped.Incontrast,bydroppinghabittechnologyfitauniquemarginof5.2%oftheeffect
size upon behavioral intention would be lost Figure 5.1 Redundancy Analysis: Effect Size
uponBehavioralIntention).Thisfindingbuildsonpreviousresearchwhichhasfoundhabit
as an important determinant of intentional behavior and still because of being conducted
basedonsinglehabitithasfoundhabitsrelativeimportancejustmoderate(seeEscobar
Rodrguez & CarvajalTrujillo 2013; Ney 2013; Pahnila, Siponen & Zheng 2011; Venkatesh,
Thong & Xu 2012), whereas multiple nonpredetermined habits represented by habit
technologyfitrelativeimportancewasfoundveryhighinthisresearch.

177

 REDUNDANCYANALYSIS:EFFECTSIZEUPONBEHAVIORALINTENTION
PE EE
0.2% 0.2%

5.2%
HTF
Figure5.1RedundancyAnalysis:EffectSizeuponBehavioralIntention (Source:Author)

Oneoftheobjectivesofthisthesisconsistedofconductingposthocmodelmodificationin
ordertoachievethebestmodelspecificationandbestfitswiththedata( =1.545, =
0.100,RMSEA=0.033,PCLOSE=0.822).Thisthesisfoundthatthebestmodelspecification
included:habittechnologyfitandsocialinfluenceasdeterminantsofbehavioralintention,
and behavioral intention as determinant of actual behavior. This section begins the
discussionwithcollateralfindingsthatreinforcetheparsimoniousspecificationthefinalHTF
Model.

Thesefindingsofredundancyoftheeffectsizesuggestthathabittechnologyfiteffectsize
could be more robust than performance expectancy and effort expectancy together in
relationshipwithbehavioralintention.Basedonthesefindings,habittechnologyfitmaybe
able to synthetize the effects of perceived usefulness and perceived ease of use from the
Technology Acceptance Model TAM (Davis, Bagozzi & Warshaw 1989), and performance
expectancyandeffortexpectancyfromUTAUT (Venkateshetal.2003).Thereasonisthat
bothscalesareinheritedfromTAM,andremainthesameinthetwomodels.

This finding may also extend the efforts to synthetize the best research traditions of
technology acceptance initiated with the Unified Theory of Acceptance and Use of
Technology (Venkatesh et al. 2003), and may constitute an original contribution to the
theoryoftechnologyacceptance.

Thisthesisdoesnotdisregardthevalueofperformanceexpectancyandeffortexpectancyin
acceptanceoftechnology.Onthecontrary,itconfirmstheirroleasimportantpredictorsof

178

behavioralintentionasitwassuggestedinVenkateshetal.(2003),andextensivelytested
thereafter,instudiessuchasMartins,OliveiraandPopovi(2014),Chu(2013),Ney(2013),
and Suksangiam and Chaiyasoonthorn (2013). By no means would this proposition imply
that ease equals habits or fit, or that usefulness equals habits or fit. Evidence provided in
this thesis (QSort) supports these conceptual differences. However, the redundancy
betweenhabittechnologyfit,performanceexpectancyandeffortexpectancymayprovidea
variable that can only be used in alternative models without performance expectancy or
effort expectancy. This may constitute an original contribution to the literature of
technologyacceptance.

Theoverlappingeffectsizeofhabittechnologyfitcouldprovidesomeempiricalsupportto
the untested hypothesis in Karahanna, Agarwal and Angst (2006) about the relationship
between workstyle compatibility and usefulness. Besides, it may confirm that habitual
behaviorisperceivedaseasy,whereasnonhabitualbehaviorisperceivedasmorecomplex,
whichhasbeenaffirmedinpreviousempiricalresearch,see(Wood,Quinn&Kashy2002)
and(Verplanken,Basetal.1998).

5.6.1 PosthocModifiedModel

Theposthocmodificationprocessrevealedthatthebestspecificationwastheoneshownin
Figure 5.2. In the final model habittechnology fit (=.44) and social influence (=.36) are
supported as determinants of behavioral intention, and behavioral intention (=.70) is
supportedasdeterminantofactualbehavioratpvalue<.001.Thefinalmodelgotabetterfit
thanthebaseandextendedmodelswithandwithoutmoderators.Italsoreachedbetter
fitthanitsownspecificationwithmoderators.Noothercombinationtriedachievedbetter
fitandprobabilisticsignificancethantheoneshowninthisfigure.

This suggests that the combination of habittechnology fit and social influence might be
better than the combination of performance expectancy, effort expectancy and social
influence or habittechnology fit, effort expectancy and social influence. Therefore it was
deemedappropriatetosuggesttheHabitTechnologyFitModelasanoriginalcontribution
tothetheoryoftechnologyacceptance.

179

 NEWMODIFIEDMODEL:HABITTECHNOLOGYFITMODEL

x df 1.545, p 0.100, GFI 0.990, CFI 0.996, RMSEA 0.033, PCLOSE 0.822.
Correlation:SIHTF=0.448***
*pvalue<.05,**pvalue<.01,***pvalue<.001
ActualBehavior(AB),BehavioralIntention(BI),HabitTechnologyFit(HTF),SocialInfluence(SI).
Figure5.2NewModifiedModel:HabitTechnologyFitModel (Source:Author)

An explanation of the findings could be found in one of the most influential theories of
human behavior, the Theory of Reasoned Action TRA (Ajzen & Fishbein 1980) detailed in
Section 2.6.2. TRA posits that attitudes and subjective norm are the most important
determinantsofbehavioralintention,whichleadstoactualbehavior.Thesimplicityofthe
models(TRAandHTF),andthefullequivalencebetweensubjectivenorm(Ajzen&Fishbein
1980) and social influence (Venkatesh et al. 2003) constructs, suggests remarkable
resemblancesbetweenTRAandthenewHTFmodel.

ThemaindifferencebetweenTRA(Ajzen&Fishbein1980)andtheHTFmodeldeveloped
inthisthesisisthatthehabittechnologyfitconstructappearsinsteadofattitudes.UTAUT
(Venkatesh et al. 2003) eliminated attitudes in the unification process, where TRA was
included. This removal of attitude has brought some of the hardest criticisms on UTAUT
(BenBoubaker&Barki2006;Yang2010;Zhang&Sun2009).However,thetheorysupports
that attitudes are contained in habits. Attitudes act as habitual responses that can be
thought as a cognitive structure, stored, and retrieved (Aarts, Verplanken & Knippenberg
1998; Petty, Fazio & Briol 2012). Extensive empirical research explains and supports this
associationofattitudesinhabits,see(Strack&Deutsch2004)and(Fazio1986).Therefore,
in the evaluation of habittechnology fit attitudes are implicitly considered. From this
perspectivetheTheoryofReasonedAction(Ajzen&Fishbein1980)wouldexplainthenew
HTFmodel.Indirectly,theHTFmodelconfirmsandextendstheTheoryofReasonedAction
(Ajzen&Fishbein1980)inanoriginalwaythatmaysignificantlycontributetotheliterature
ofbehavior,habit,personenvironmentfit,andtechnologyacceptance.

180

5.7 Summary

Inthissectionthemainfindingsofthisresearchhavebeendiscussed.Apositiverelationship
between habittechnology fit and behavioral intention was found in the results of this
thesis.Ageandexperiencewerealsofoundtomoderatetheirrelationship.Thesefindings
havebeenrelatedtotheliteratureandtheory,remarkingthecontributionsofthisresearch.
Findings about the base, extended and modified models, investigated in this thesis, were
discussedaswell.Thefollowingsectionprovidesconclusiontothisthesis.

181

 CHAPTER6 CONCLUSIONS
6.1 Objective

The purpose of this chapter is to report the contributions and the implications of the key
findings. The limitations of this study are acknowledged, and recommendations for future
researcharepresented.

6.2 Overviewofthesisobjectivesandresearchquestions

Thisthesishadthreemainobjectives:tostudytherelationshipofthestructureofhabitsand
behavioral intention in individuals, through habittechnology fit; to analyze the effect of
includinghabittechnologyfit,asanewconstruct,intheUnifiedTheoryofAcceptanceand
Use of Technology (UTAUT) framework; and to conduct posthoc model modification in
attempttoimprovetheresearchmodel.Thesemaingoalswereachieved.

Three research questions were formulated and answered in this thesis. In regards to the
firstresearchquestiontheimpactofhabittechnologyfituponbehavioralintentioninthe
context of technology acceptance was positive and significant. Age and experience
moderatedthisrelationshipmakingitstrongerforolderandmoreexperiencedindividuals.
However,thesemoderatorswereunlikelytoachievemodelfitinothersamplesofthesame
population.

Forthesecondquestion,theeffectsofincludinghabittechnologyfitintheUnifiedTheory
of Acceptance and Use of Technology model had several effects at different times of this
research.ByincludinghabittechnologyfitinUTAUTtheexplanationofbehavioralintention
improves. However, the new construct makes performance expectancy and effort
expectancys shared variance redundant. The effect of including habittechnology fit in
UTAUT pushes the limits of the analysis techniques, and forces a decision between habit
technology fit or performance expectancy and whether effort expectancy should be
includedinthemodel.

Finally,thethirdquestionwasansweredinregardtowhetheritwaspossibletoimprovethe
model specification of the Unified Theory of Acceptance and Use of Technology by
182

conducting posthoc model modification. Performance expectancy and effort expectancy

werereplacedwithhabittechnologyfit,becausethelossoftheeffectsizeuponbehavioral
intention was greater in the other way (.4% vs 5.2% loss). The moderators were also
dropped, and this allowed achievement of optimal model fit, but also probabilistic
significance( =1.545, =0.100,RMSEA=0.033,PCLOSE=0.822).Byperformingthe
mentionedmodificationsUTAUTwasimprovedandanewHabitTechnologyFitmodelwas
proposed.

6.3 Keytheoreticalcontributions

Themaincontributionofthisresearchisthatithasconceptualizedanewconstructhabit
technology fitand empirically investigated its relationship with behavioral intention.
Habits, other than the one which corresponds with the target behavior, had rarely been
considered, if ever, in the theory of habit or technology acceptance. This thesis found
theoretical grounds to suggest that habits cannot exist in a pure isolated form, and their
complete structure determines taste, choice and understanding (Bourdieu 1984; Swartz
2002;Wozniak2009).Itfoundempiricalevidencethatmultiplehabitspredeterminedby
theresearcherhadanimpactuponbehavioralintention.Italsofoundliteratureofperson
environmentfit,suchas(Cable&DeRue2002;KristofBrown,Zimmerman&Johnson2005),
assertingthatthesalientcharacteristicswhichdeterminebehavioralintentionvaryamong
individuals, and perceived fit may capture these characteristics which are salient to
individualdynamically.Therefore,thecombinationofperceivedfitandhabitscouldpossibly
anticipate that the better the fit between habits and the technology, the higher the
intention to use it. This concept was empirically confirmed. Thus, the theorized habit
technology fit construct constitutes an original contribution to the theory of habit and
technologyacceptance.

Second, habittechnology fit operationalization may open a new avenue for research in
whichthestructureofhabitsisconsidered,measuringonlysingleorpredeterminedhabits.
This operationalization is different from previous approaches that measure single
predetermined: (EscobarRodrguez & CarvajalTrujillo 2013; Limayem & Cheung 2008);
single semipredetermined: (Klckner, Matthies & Hunecke 2003); single non
183

predetermined: (Verplanken & Orbell 2003); multiple predetermined: (Chen & Lai 2011;
Klckner, Matthies & Hunecke 2003; Landis, Triandis & Adamopoulos 1978); and multiple
semipredetermined: (Bamberg & Schmidt 2003; Klckner, Matthies & Hunecke 2003;
Verplanken,Basetal.1998).Themaindifferenceisthenumberandsalienceofthehabits
consideredfromtheindividualsperspective.

Third, this thesis contributes to the theory of technology acceptance by presenting a new
model. The HabitTechnology Fit model (see Figure 6.1) is parsimonious, and as a whole
modelitovercameUTAUTbyachievingabetterfitwithdata,butalsoacceptablestatistical
significanceasamodel.Thismodelisanoriginalcontributionofthisthesistotheliterature
ofhabit,andtechnologyacceptance.

HABITTECHNOLOGYFITMODEL

Figure6.1HabitTechnologyFitModel (Source:Author)

6.4 Researchimplications

This thesis uncovered and addressed a gap in the literature of habits where a whole
structure of habits has been overlooked in its relationship with intention and behavior by
onlypayingattentiontosinglehabits.However,thefindingsofthisresearchmaydrivethe
attentionofresearcherstomanyaspectsandrelationshipsoftheotherhabits.

184

Habittechnologyfitwasproposedasanewconstructintechnologyacceptance.However,
researchers may continue to find apparent differentiation concerns with performance
expectancy and high collinearity with effort expectancy. Therefore, an implication of the
newconstructmightbethatthesefactorsmaynotbeabletobetestedtogethermostof
the time without risking the validity of the results by multiple regression and structural
equationmodelingtechniques.

Inthecontextofthisresearch,itwasfoundthathabittechnologyfitandsocialinfluenceare
the best predictors of behavioral intention. Habittechnology fit contained almost all the
explanatorypowerofperformanceexpectancyandeffortexpectancy.Thismayimplythat
researchers could use habittechnology fit in substitution of these constructs, but also of
equivalentconstructssuchasperceivedusefulnessandperceivedeaseofuse(Davis1989),
andyetmaintaintheexplanatorypoweruponintention.

Furthermore,thenewmodelpresentedmightbethebeginningofadifferentlineofmodels
andextensions.However,itmayalsobringtheattentionofresearcherstotheimportance
of the probabilistic aspects of model fit in structural equation modeling as an indicator of
significanceforthemodelsasawhole.

6.5 Practicalimplications

Apracticalimplicationofthisresearchisconnectedwiththeindustryoftheinformationand
communicationtechnologies.Inordertoreduceuserresistancetochange,amajorfactorof
failureinadoptionoftechnology(doCantoCavalheiro&Joia2013;Jiang,Muhanna&Klein
2000),newproductdevelopersanddesignersmayneedtoconsiderdevelopingtechnology
which is compatible with the habits of their users. This study identified that habit
technology fit positively influences behavioral intention. It means that the better the fit
betweenapersonshabitsandthetechnology,thehighertheintentionstoacceptanduse
the proposed technology. Previous empirical research has shown that radically innovative
products may fail because the design characteristics of the new product are incongruent
withtheexisting,withgreaterlearningcostsfortheuseridentifiedasthereasonoffailure

185

(Mugge&Dahl2013).Itisnotthatradicalinnovationsarenotneeded;thesefindingsmean
thatithastobedesignedtobecompatiblewiththecurrenthabitsofitsusers.

Utilization of the Internet is growing at an astonishing rate, (Chung & Tan 2004) and
particularlySoftwareasaServiceindustryexpectscontinuedfastgrowthinthenextyears
(17.5% annually until 2016). Since SaaS is offered ondemand, the focus of the industry is
largely moving from CIOs to endusers, who must be pleased in order to keep corporate
sales and grow (PwC 2013). This thesis points to an important aspect of individual
acceptance and use of technology. For individuals to accept technology, it has to be
designedcompatiblewiththeirhabits.

Companiesarefrequentlyfocusedoninnovation.Butbyconcentratingoninnovationonly,
high costs of learning can are imposed on the enduser (Mugge & Dahl 2013). When the
user has several options, software that is less compatible with the user habits may face
serioustrouble.

Thesethesisfindingsmayalsohaveimplicationsforchangesandinterventionsdemanding
to move away from current habits and adopt incompatible behaviors. Requesting drastic
changesinbehaviorisanunlikelysuccessstrategyaccordingtothefindingsofthisresearch.
Instead, progressive changes, which are more compatible with the individuals habits, are
morelikelytobeacceptedandsustained.Progressivechangesmayreshapeprevioushabits
andeventuallyleadtothetargetbehavior.Habitsareelasticandplastic,sotheychangebut
changeslowlyastheypreservepreciousexperience.Then,theysustainpatternsofbehavior
(Hodgson2010;Verplanken&Wood2006;Wood&Neal2009).

6.6 LimitationsandFutureResearch

Asineveryfield,thelongtermaspirationofthetechnologyacceptanceresearchmightbe
finding a robust, and yet parsimonious theory. It is important for a theory of technology
acceptancetobecapableofpredictingacceptanceacrosssettingsandtechnologies.Oneof
the limitations of this research is the level at which its findings can be generalized. The
targetpopulationofthisstudywasdefinedascomprisingadultswhoareSoftwareasa

186

Service users in public clouds, and understand Englishwas approached with a non
probability sample. Nonprobability samples are not ideal, yet sometimes necessary when
the elements of the population cannot be identified, and therefore cannot be randomly
selected (Blaikie 2010). However, this limitation was addressed by using a Respondent
DrivenSampling technique which reduces the bias by homophily (Heckathorn 2002). This
researchalsoconsideredinthesampletherepresentationofcharacteristicsthathavebeen
identified as influential in technology acceptance such as age, gender, experience,
voluntariness,hedonicandutilitarianutilization,andculturalindexofindividualism.

Whereasanindividualcanretrospectivelybecomeawareofhishabitsbythetracesofhis
unaware actions (Mittal 1988), selfreport bias that comes from a seminatural setting
remainsalimitation.Whenpeopleareaskedtoreportonthemselves,theremightbeagap
betweenthereportandreality(Blaikie2010).Thisisalimitationcommontomostresearch
intechnologyacceptance,see(Venkateshetal.2003),(Davis1989),and(Fishbein&Ajzen
1975).Perceivedfitasaselfreportmeasure,maynotallowestimationoftheindependent
effectsofthepersonscharacteristics,apartfromtheenvironments(Edwards1991,1996).
However,perceivedfithasprovedbettertopredictintentionandbehaviorthanobjective
measures of fit (Cable & DeRue 2002; KristofBrown & Billsberry 2012; KristofBrown,
Zimmerman&Johnson2005).Inthefuture,researchersmayattempttoseparatetheeffect
ofspecifichabitsfromhabittechnologyfitwithredundancyanalysis.

The new HabitTechnology Model presented in this thesis needs further testing across
contexts and technologies. In doing this, it is particularly important to apply confirmatory
techniquessuchascovariancebasedstructuralequationmodelingtostrengthenorweaken
the general validity of the model. Before extending or modifying the model, it is highly
recommended to test and report all the model fit indicators. Probabilistic values for chi
squareandRMSEAareparticularlyimportantfortheoreticalconfirmationpurposes(Byrne
2010;Hairetal.2010).Moreresearchisalsoneededtoestablishtheroleofcultureacross
several countries, and longitudinal studies using the HabitTechnology Fit Model are
recommended.

187

OneoftheconstructswhichisconsideredadeterminantofactualbehaviorinUTAUT,could
not be tested. In the particular context of this research, facilitating conditions failed to
achieve convergent and discriminant validity. The construct was dropped, and the model
hadtobetestedwithoutit.Furtherresearchisneededtoimprovethemeasurementscale
of facilitating conditions. This construct should either be defined as onedimensional and
made more general, or acknowledged as multidimensional and develop redundancy items
foreachdimension.Thisthesisprovidedimportantresultsaboutthetheoreticalvalidityof
UTAUT,butnotbeingabletotestfacilitatingconditionsinthemodelisalimitationofthis
research.Thus,replicatingtheoriginalspecificationofUTAUTbyconfirmatorytechniques
remainsanimportanttaskforfuturework.

Measurement of habits has been approached with habittechnology fit. However, this
approach is a proxy to measure habits, not the measurement of habits per se. Future
researchisencouragedtodefinehabitswithanempiricalandontologicalapproach.Future
success and significance of habits in advancing to sound theory may depend on clearly
understandinghabitsidentity,essence,unity,anddependence(seeWelty&Guarino2001).

188

6.7 Summary

This chapter provided closure to the thesis. Research objectives and research questions
were revisited and answered from the findings. It also presented the key contributions of
this research, such as the conceptualization of habittechnology fit construct, empirical
evidenceofitspositiverelationshipwithbehavioralintention,andthecontributionofanew
model of technology acceptance based on habittechnology fit and social influence as
determinants of intention. This chapter discussed research and practical implications,
acknowledgedlimitationsandsuggestedfutureresearch.

189

REFERENCES

Aarts,H&Dijksterhuis,A2000,'TheAutomaticActivationofGoalDirectedBehaviour:The
CaseofTravelHabit',JournalofEnvironmentalPsychology,vol.20,no.1,pp.7582.
Aarts, H & Dijksterhuis, A 2000, 'Habits as Knowledge Structures: Automaticity in Goal
Directed Behavior', Journal of Personality and Social Psychology, vol. 78, no. 1, pp.
5363.
Aarts, H, Paulussen, T & Schaalma, H 1997, 'Physical Exercise Habit: On the
Conceptualization and Formation of Habitual Health Behaviours', Health education
research,vol.12,no.3,pp.36374.
Aarts, H, Verplanken, B & Knippenberg, A 1998, 'Predicting Behavior from Actions in the
Past: Repeated Decision Making or a Matter of Habit?', Journal of Applied Social
Psychology,vol.28,no.15,pp.135574.
Abelson, RP 1981, 'Psychological Status of the Script Concept', American Psychologist, vol.
36,no.7,pp.71529.
Abowitz,KK2011,'AchievingPublicSchools',EducationalTheory,vol.61,no.4,pp.46789.
Abraham,C&Sheeran,P2003,'ImplicationsofGoalTheoriesfortheTheoriesofReasoned
ActionandPlannedBehaviour',CurrentPsychology,vol.22,no.3,pp.26480.
AbuShanab, E & Pearson, JM 2007, 'Internet Banking in Jordan: The Unified Theory of
Acceptance and Use of Technology (UTAUT) Perspective', Journal of Systems and
InformationTechnology,vol.9,no.1,pp.7897.
AbuShanab, E, Pearson, JM & Setterstrom, AJ 2010, 'Internet Banking and Customers'
Acceptance in Jordan: The Unified Model's Perspective', Communication of the
AssociationforInformationSystems,vol.26,no.23,pp.493524.
Ahmad, KZB 2011, 'Group Size as a Moderator of the Effect of Equity Sensitivity on
EmployeeJobSatisfaction',InternationalJournalofManagement,vol.28,no.3,pp.
71629.
Ajzen, I 1987, 'Attitudes, Traits, and Actions: Dispositional Prediction of Behavior in
PersonalityandSocialPsychology',Advancesinexperimentalsocialpsychology,vol.
20,pp.163.
1991, 'The Theory of Planned Behavior', Organizational Behavior and Human Decision
Processes,vol.50,no.2,pp.179211.
Ajzen,I2002,'ResidualEffectsofPastonLaterBehavior:HabituationandReasonedAction
Perspectives',PersonalityandSocialPsychologyReview,vol.6,no.2,pp.10722.
Ajzen, I & Fishbein, M 1980, Understanding Attitudes and Predicting Social Behavior,
PrenticeHall,EnglewoodCliffs,NJ.
AlGahtani, SS, Hubona, GS & Wang, J 2007, 'Information Technology (It) in Saudi Arabia:
CultureandtheAcceptanceandUseofIt',Information&Management,vol.44,no.
8,pp.68191.
AlHtaybat,KA&vonAlbertiAlhtaybat,L2013,'MobilityinCorporateFinancialReporting:
The Unified Theory of Acceptance and Use of Technology Model', Life Science
Journal,vol.10,no.4.

Alakrpp, I, Valtonen, A, Alakulju, H & Hrm, H 2010, 'Acceptance of Practices: Case

Bioactive Innovations in Health Care Market', International Journal of Electronic
BusinessManagement,vol.8,no.4,pp.30411.
Allen,KM2010,'HabitinGeorgeEliot'stheMillontheFloss',StudiesinEnglishLiterature,
15001900,vol.50,no.4,pp.83152.
Alshuwaikhat, HM & Nkwenti, DI 2003, 'Collaborative Planning and Management
Frameworks: Approaches to Effective Urban Governance by Adoption of Emerging
Technologies',InternationalJournalofManagement,vol.20,no.3,pp.294305.
Annelies,EMvV,DePater,IE&FloorVan,D2007,'WorkValueFitandTurnoverIntention:
SameSourceorDifferentSourceFit',JournalofManagerialPsychology,vol.22,no.
2,pp.188202.
Anonymous2009a,'LearningandLeadingwithHabitsofMind:16EssentialCharacteristics
forSuccess',Adolescence,vol.44,no.173,pp.245.
2009b,'TimeDrills:DeepTimeRapidTimeExercises',ArtJournal,vol.68,no.4,pp.44
61.
Anshel, MH & Kang, M 2007, 'Effect of an Intervention on ReplacingNegative Habits with
PositiveRoutinesforImprovingFullEngagementatWork:ATestoftheDisconnected
ValuesModel',ConsultingPsychologyJournal:PracticeandResearch,vol.59,no.2,
pp.11025.
Anshel, MHP & Kang, M 2007, 'An OutcomeBased Action Study on Changes in Fitness,
BloodLipids,andExerciseAdherence,UsingtheDisconnectedValues(Intervention)
Model',BehavioralMedicine,vol.33,no.3,pp.85100.
Arbuckle,JL2010,'IbmSpssAmos19UsersGuide',Crawfordville,FL:AmosDevelopment
Corporation,p.635.
Arbuthnott, KD 2009, 'Education for Sustainable Development Beyond Attitude Change',
InternationalJournalofSustainabilityinHigherEducation,vol.10,no.2,pp.15263.
Archer,MS2010,'Routine,Reflexivity,andRealism*',SociologicalTheory,vol.28,no.3,pp.
272303,54.
Aristotle1976,Ethics,Penguin,Harmondsword.
Armstrong,EG1986,'UniformNumbers',TheAmericanJournalofSemiotics,vol.4,no.1/2,
pp.99108,1020,2227.
Avital, M & Te'eni, D 2009, 'From Generative Fit to Generative Capacity: Exploring an
Emerging Dimension of Information Systems Design and Task Performance',
InformationSystemsJournal,vol.19,no.4,pp.34567.
Aydin,C2009,'OntheSignificanceofIdeals:CharlesS.PeirceandtheGoodLife',CharlesS.
PeirceSociety.TransactionsoftheCharlesS.PeirceSociety,vol.45,no.3,pp.42243.
Babbie,E2013,TheBasicsofSocialResearch,CengageLearning.
Bacon, DR, Sauer, PL & Young, M 1995, 'Composite Reliability in Structural Equations
Modeling',EducationalandPsychologicalMeasurement,vol.55,no.3,pp.394406.
Bagozzi, RP 1981, 'Attitudes, Intentions, and Behavior: A Test of Some Key Hypotheses',
Journal of Personality and Social Psychology; Journal of Personality and Social
Psychology,vol.41,no.4,p.607.
Bagozzi,RP&Warshaw,PR1990,'TryingtoConsume',Journalofconsumerresearch,pp.
12740.
Baloh,P2007,'TheRoleofFitinKnowledgeManagementSystems:TentativePropositions
oftheKmsDesign',JournalofEndUserComputing,vol.19,no.4,pp.22.
191

Bamberg,S&Schmidt,P2003,'Incentives,Morality,orHabit?PredictingStudentsCarUse
forUniversityRouteswiththeModelsofAjzen,Schwartz,andTriandis',Environment
andbehavior,vol.35,no.2,pp.26485.
Bandura, A 1986, 'Social Foundations of Thought and Action: A Social Cognitive Theory',
Changes,vol.1,no.US:PrenticeHall.
Bandyopadhyay, K & Fraccastoro, KA 2007, 'The Effect of Culture on User Acceptance of
Information Technology', Communications of the Association for Information
Systems,vol.19,no.1,pp.23.
Bansal,G2011,'EBookUsage:RoleofEnvironmentalConsciousness,PersonalityandPast
Usage',TheJournalofComputerInformationSystems,vol.52,no.2,pp.93104.
Bargh, JA 1989, 'Conditional Automaticity: Varieties of Automatic Influence in Social
PerceptionandCognition',Unintendedthought,vol.3,pp.5169.
Bargh, JA 1990, 'AutoMotives: Preconscious Determinants of Social Interaction', in
Handbook of Motivation and Cognition: Foundations of Social Behavior, Guilford
Press,NewYork,NY,vol.2,pp.93130.
2002, 'Losing Consciousness: Automatic Influences on Consumer Judgment, Behavior,
andMotivation',Journalofconsumerresearch,vol.29,no.2,pp.2805.
Bargh, JA, Gollwitzer, PM, LeeChai, A, Barndollar, K & Trtschel, R 2001, 'The Automated
Will:NonconsciousActivationandPursuitofBehavioralGoals',JournalofPersonality
andSocialPsychology;JournalofPersonalityandSocialPsychology,vol.81,no.6,p.
1014.
Barnes, SJ 2011, 'Understanding Use Continuance in Virtual Worlds: Empirical Test of a
ResearchModel',Information&Management,vol.48,no.8,pp.3139.
Bartlett,FC1997,Remembering:AStudyinExperimentalandSocialPsychology,Cambridge
UniversityPress,NewYork.NY.
Bayton, JA 1957, 'Motivation, Cognition, LearningBasic Factors in Consumer Behavior',
JournalofMarketing(pre1986),vol.22,no.1,p.282.
Beck, U 2004, Risk Society: Towards a New Modernity, vol. 17, Sage Publications Limited,
London.
Becker, GM, DeGroot, MH & Marschak, J 1963, 'Stochastic Models of Choice Behavior',
BehavioralScience,vol.8,no.1,pp.4155.
Becker,GS1992,'Habits,Addictions,andTraditions',Kyklos,vol.45,no.3,pp.32745.
Becker,MC2005,'AFrameworkforApplyingOrganizationalRoutinesinEmpiricalResearch:
Linking Antecedents, Characteristics and Performance Outcomes of Recurrent
InteractionPatterns',IndustrialandCorporateChange,vol.14,no.5,pp.81746.
Ben Boubaker, K & Barki, H 2006, 'A Decomposed Model of It ArtifactRelated Beliefs as
AntecedentsofItAcceptanceandUse'.
Berardelli,A,Curr,A,Fabbrini,G,Gilio,F&Manfredi,M2003,'PathophysiologyofTicsand
TouretteSyndrome',JournalofNeurology,vol.250,no.7,pp.7817.
Bergeron, F, Raymond, L, Rivard, S & Gara, MF 1995, 'Determinants of Eis Use: Testing a
BehavioralModel',DecisionSupportSystems,vol.14,no.2,pp.13146.
Bergvik,S,Svendsen,GB&Evjemo,B2006,'BeyondUtility.ModellingUserAcceptanceof
MobileDevices'.
Berk,G&Galvan,D2009,'HowPeopleExperienceandChangeInstitutions:AFieldGuideto
CreativeSyncretism',TheoryandSociety,vol.38,no.6,pp.54380.

192

Berridge, KC 2007, 'The Debate over Dopamine's Role in Reward: The Case for Incentive
Salience',Psychopharmacology,vol.191,no.3,pp.391431.
Biesta, G 2007, 'Why "What Works" Won't Work: EvidenceBased Practice and the
DemocraticDeficitinEducationalResearch',EducationalTheory,vol.57,no.1,pp.1
22.
Bijker,WE,Hughes,TP&Pinch,TJ1987,TheSocialConstructionofTechnologicalSystems:
New Directions in the Sociology and History of Technology, MIT Press, Cambridge,
Mass.
Blaikie,N2010,DesigningSocialResearch,2ndedn,PolityPress,Cambridge.
Bonne,K,Vermeir,I,BergeaudBlackler,F&Verbeke,W2007,'DeterminantsofHalalMeat
ConsumptioninFrance',BritishFoodJournal,vol.109,no.5,pp.36786.
Bourdieu, P 1984, Distinction: A Social Critique of the Judgement of Taste, Harvard
UniversityPress,Cambridge,Mass.
Bourdieu,P1985,'TheSocialSpaceandtheGenesisofGroups',TheoryandSociety,vol.14,
no.6,pp.72344.
Bourdieu,P1990,TheLogicofPractice,StanfordUniversityPress,Stanford,Calif.
Bourdieu,P2008,The Bachelors'Ball:TheCrisisofPeasantSocietyin Barn,Universityof
ChicagoPress,Chicago.
Bourdieu,P&Wacquant,LJ1992,AnInvitationtoReflexiveSociology,UniversityofChicago
Press,Chicago.
Bover, O 1991, 'Relaxing Intertemporal Separability: A Rational Habits Model of Labor
SupplyEstimatedfromPanelData',JournalofLaborEconomics,pp.85100.
Boyer, P & Linard, P 2006, 'Why Ritualized Behavior? Precaution Systems and Action
Parsing in Developmental, Pathological and Cultural Rituals', Behavioral and Brain
Sciences,vol.29,no.6,pp.595613;discussion50.
Brand, M 2003, 'Market Intelligence: Making Market Research Work for Microfinance',
Accion'sInSightSeries,no.7.
Brand,W2009,'Hume'sAccountofCuriosityandMotivation',JournalofValueInquiry,vol.
43,no.1,pp.8396.
Brinkmann, S 2007, 'Culture as Practices: A Pragmatist Conception', Journal of Theoretical
andPhilosophicalPsychology,vol.2728,no.21,pp.192212.
Brockelman,PT2002,'HabitsandPersonalGrowth:TheArtofthePossible',OTJR,vol.22,
pp.1830.
Brder,A&Schiffer,S2006,'AdaptiveFlexibilityandMaladaptiveRoutinesinSelectingFast
and Frugal Decision Strategies', Journal of Experimental Psychology: Learning,
Memory,andCognition,vol.32,no.4,pp.90418.
Brown,SR1997,'TheHistoryandPrinciplesofQMethodologyinPsychologyandtheSocial
Sciences',paperpresentedtoRedattheBritishPsychologicalSocietySymposiumon,
AQuestforaScienceofSubjectivity:TheLifeworkofWilliamStephenson,University
ofLondon.
Browne,MW,Cudeck,R,Bollen,KA&Long,JS1993,'AlternativeWaysofAssessingModel
Fit',inKABollen&JSLong(eds),TestingStructuralEquationModels,SAGE,Newbury
Park,CA,pp.13662.
Bryman,A&Bell,E2007,BusinessResearchMethods,OxfordUniversityPress,USA.
Bula,G2004,'ObervationsontheDevelopmentofCyberneticIdeasinColombia:ATribute
toStaffordBeer',Kybernetes,vol.33,no.3/4,pp.64758.
193

Buxton,W1986,'TheresMoretoInteractionThanMeetstheEye:SomeIssuesinManual
Input', User centered system design: New perspectives on humancomputer
interaction,pp.31937.
Byrne,BM2010,StructuralEquationModelingwithAmos:BasicConcepts,Applications,and
Programming,2ndedn,Routledge,NewYork,NY.
Byrne,D2002,InterpretingQuantitativeData,Sage,London.
Cable, DM & DeRue, DS 2002, 'The Convergent and Discriminant Validity of Subjective Fit
Perceptions',JournalofAppliedPsychology,vol.87,no.5,pp.875.
Cable, DM & DeRue, DS 2002, 'The Convergent and Discriminant Validity of Subjective Fit
Perceptions',JournalofAppliedPsychology,vol.87,no.5,pp.87584.
Cable, DM & Judge, TA 1996, 'PersonOrganization Fit, Job Choice Decisions, and
Organizational Entry', Organizational Behavior and Human Decision Processes, vol.
67,no.3,pp.294311.
Canin, L, Dolcini, MM & Adler, NE 1999, 'Barriers to and Facilitators of HivStd Behavior
Change: Intrapersonal and RelationshipBased Factors', Review of General
Psychology,vol.3,no.4,pp.33871.
Caplan, RD 1987, 'PersonEnvironment Fit Theory and Organizations: Commensurate
Dimensions, Time Perspectives, and Mechanisms', Journal of Vocational Behavior,
vol.31,no.3,pp.24867.
Carless, SA 2005, 'PersonJob Fit Versus PersonOrganization Fit as Predictors of
Organizational Attraction and Job Acceptance Intentions: A Longitudinal Study',
JournalofOccupationalandOrganizationalPsychology,vol.78,pp.41129.
Carmines,EG&McIver,JP1981,'AnalyzingModelswithUnobservedVariables:Analysisof
CovarianceStructures',Socialmeasurement:Currentissues,pp.65115.
Carter,J&Fourney,D2004,'UsingaUniversalAccessReferenceModeltoIdentifyFurther
GuidanceThatBelongsinIso16071',UniversalAccessintheInformationSociety,vol.
3,no.1,pp.1729.
Charmaz,K2002,'TheSelfasHabit:TheReconstructionofSelfinChronicIllness',OTJR,vol.
22,pp.3141.
Charng, HW, Piliavin, JA & Callero, PL 1988, 'Role Identity and Reasoned Action in the
PredictionofRepeatedBehavior',SocialPsychologyQuarterly,pp.30317.
Chatterjee, P 2010, 'EService Brand Extensions: The Role of Perceived Fit and Category
UsageLevelonAdoption',JournalofServiceScience,vol.3,no.1,pp.713.
Chen,CF&Chao,WH2010,'HabitualorReasoned?UsingtheTheoryofPlannedBehavior,
Technology Acceptance Model, and Habit to Examine Switching Intentions toward
PublicTransit',TransportationResearchPartF:TrafficPsychologyandBehaviour,vol.
14,no.2,pp.12837.
Chen, CF & Lai, WT 2011, 'The Effects of Rational and Habitual Factors on Mode Choice
Behaviors in a MotorcycleDependent Region: Evidence from Taiwan', Transport
Policy,vol.18,no.5,pp.7118.
Chen, JL 2011, 'The Effects of Education Compatibilityand Technological Expectancy on E
LearningAcceptance',Computers&Education,vol.57,no.2,pp.150111.
Chin, WW 1995, 'Partial Least Squares Is to Lisrel as Principal Components Analysis Is to
CommonFactorAnalysis',TechnologyStudies,vol.2,no.2,pp.3159.
Chu, KM 2013, 'Motives for Participation in Internet Innovation Intermediary Platforms',
InformationProcessing&Management,vol.49,no.4,pp.94553.
194

Chung, J & Tan, FB 2004, 'Antecedents of Perceived Playfulness: An Exploratory Study on

User Acceptance of General InformationSearching Websites', Information &
Management,vol.41,no.7,pp.86981.
Collet, F 2009, 'Does Habitus Matter? A Comparative Review of Bourdieu's Habitus and
Simon's Bounded Rationality with Some Implications for Economic Sociology',
SociologicalTheory,vol.27,no.4,pp.41934.
Compeau,DR&Higgins,CA1995,'ComputerSelfEfficacy:DevelopmentofaMeasureand
InitialTest',MISQuarterly,pp.189211.
Comtet,H2013,'Acceptanceof3dGesturesBasedonAge,GenderandExperience',Gjovik
University.
Cornacchia,M,Baroncini,V&Livi,S2008,'PredictingtheInfluenceofEmergingInformation
and Communication Technologies on Home Life', in Web Information Systems and
Technologies,Springer,pp.184200.
Crepeau,E2007,'Introduction',OTJR,vol.27,pp.35.
Crissman,P1942,'ThePsychologyofJohnDewey',PsychologicalReview,vol.49,no.5,pp.
44162.
Cronbach, LJ 1951, 'Coefficient Alpha and the Internal Structure of Tests', Psychometrika,
vol.16,no.3,pp.297334.
Cronbach, LJ & Meehl, PE 1955, 'Construct Validity in Psychological Tests', Psychological
Bulletin,vol.52,no.4,p.281.
Crossley,N2013,'HabitandHabitus',Body&Society,vol.19,no.23,pp.13661.
Crowther,D&Lancaster,G2012,ResearchMethods:AConciseIntroductiontoResearchin
ManagementandBusinessConsultancy,2ndedn,TaylorandFrancis,NewYork.
Cutchin, MP 2000, 'Retention of Rural Physicians: Place Integration and the Triumph of
Habit',OTJR,vol.20,pp.10611.
2007, 'From Society to Self (and Back) through Place: Habit in Transactional Context',
OTJR,vol.27,pp.509.
Dabholkar,PA&Bagozzi,RP2002,'AnAttitudinalModelofTechnologyBasedSelfService:
Moderating Effects of Consumer Traits and Situational Factors', Journal of the
AcademyofMarketingScience,vol.30,no.3,pp.184201.
Danner, U, Aarts, H, Papies, E & de Vries, N 2011, 'Paving the Path for Habit Change:
Cognitive Shielding of Intentions against Habit Intrusion', British journal of health
psychology,vol.16,no.Pt1,p.189.
Danner,UN,Aarts,H& deVries,NK2008,'HabitVs.Intentioninthe PredictionofFuture
Behaviour:TheRoleofFrequency,ContextStabilityandMentalAccessibilityofPast
Behaviour',BritishJournalofSocialPsychology,vol.47,pp.24565.
Davis, FD 1986, 'A Technology Acceptance Model for Empirically Testing New EndUser
Information Systems: Theory and Results', PhD thesis, Massachusetts Institute of
Technology.
Davis, FD 1989, 'Perceived Usefulness, Perceived Ease of Use, and User Acceptance of
InformationTechnology',MISQuarterly,vol.13,no.3,pp.31940.
Davis,FD,Bagozzi,RP&Warshaw,PR1989,'UserAcceptanceofComputerTechnology:A
ComparisonofTwoTheoreticalModels',ManagementScience,pp.9821003.
Davis, FD, Bagozzi, RP & Warshaw, PR 1992, 'Extrinsic and Intrinsic Motivation to Use
Computers in the Workplace', Journal of Applied Social Psychology, vol. 22, no. 14,
pp.111132.
195

Davis, FD & Venkatesh, V 2004, 'Toward Preprototype User Acceptance Testing of New
Information Systems: Implications for Software Project Management', Engineering
Management,IEEETransactionson,vol.51,no.1,pp.3146.
Davis,JT2001,'GonebutNotForgotten:DeclarativeandNondeclarativeMemoryProcesses
andTheirContributionstoResilience',BulletinoftheMenningerClinic,vol.65,no.4,
pp.45170.
Dawis,RV1992,'PersonEnvironmentfitandJobSatisfaction',inCJCranny,PCSmith&EF
Stone (eds), Job Satisfaction: How People Feel About Their Jobs and How It Affects
TheirPerformance,LexingtonBooks,p.326.
De Bruijn, G & Rhodes, R 2010, 'Exploring Exercise Behavior, Intention and Habit Strength
Relationships',ScandinavianJournalofMedicine&ScienceinSports,vol.21,no.3,
pp.48291.
DeBruijn,GJ,Kremers,SPJ,DeVet,EE,DeNooijer,J,VanMechelen,W&Brug,J2007,'Does
Habit Strength Moderate the IntentionBehaviour Relationship in the Theory of
PlannedBehaviour?TheCaseofFruitConsumption',PsychologyandHealth,vol.22,
no.8,pp.899916.
de Bruijn, GJ, Kremers, SPJ, Singh, A, van den Putte, B & van Mechelen, W 2009, 'Adult
Active Transportation:: Adding Habit Strength to the Theory of Planned Behavior',
Americanjournalofpreventivemedicine,vol.36,no.3,pp.18994.
De Bruijn, GJ & Van Den Putte, B 2009, 'Adolescent Soft Drink Consumption, Television
ViewingandHabitStrength.InvestigatingClusteringEffectsintheTheoryofPlanned
Behaviour',Appetite,vol.53,no.1,pp.6675.
de Nooijer, J, Onnink, M & van Assema, P 2010, 'Vitamin D Supplementation in Young
Children: Associations with Theory of Planned Behaviour Variables, Descriptive
Norms,MoralNormsandHabits',Publichealthnutrition,vol.13,no.08,pp.127985.
DeLone, WH & McLean, ER 1992, 'Information Systems Success: The Quest for the
DependentVariable',INFORMATIONSYSTEMSRESEARCH,vol.3,no.1,pp.6095.
Dennis, AR, Wixom, BH & Vandenberg, RJ 2001, 'Understanding Fit and Appropriation
EffectsinGroupSupportSystemsViaMetaAnalysis',MISQuarterly,pp.16793.
DeVellis, RF 2012, Scale Development: Theory and Applications, 3rd edn, vol. 26, Sage
Publications,Inc,ThousandOaks,Calif.
Dewey,J2002,HumanNatureandConduct(Reprint,1922),DoverPub.Co.,Mineola,New
York.
Dishaw, MT & Strong, DM 1999, 'Extending the Technology Acceptance Model with Task
TechnologyFitConstructs',Information&Management,vol.36,no.1,pp.921.
DiStefano, C, Zhu, M & Mindrila, D 2009, 'Understanding and Using Factor Scores:
Considerations for the Applied Researcher', Practical Assessment, Research &
Evaluation,vol.14,no.20,pp.111.
do Canto Cavalheiro, GM & Joia, LA 2013, 'Examining Sources of Resistance to the
ImplementationofaPatentManagementSysteminaDevelopingCountry:Evidence
fromaCaseStudyoftheBrazilianPatentOffice',inElectronicGovernment,Springer,
pp.26273.
DobbsAllsopp, FW 2005, 'The Delight of Beauty and Song of Songs 4:17', Interpretation,
vol.59,no.3,pp.26077.
Donham, J 2010, 'Creating Personal Learning through SelfAssessment', Teacher Librarian,
vol.37,no.3,pp.1421.
196

Donley,AM2012,ResearchMethods,InfobasePublishing.
Dunn,WW2000,'Habit:What'stheBrainGottoDowithIt?',OTJR,vol.20,pp.620.
Duyck, P, Pynoo, B, Devolder, P, Voet, T, Adang, L, Ovaere, D & Vercruysse, J 2010,
'Monitoring the Pacs Implementation Process in a Large University Hospital
DiscrepanciesbetweenRadiologistsandPhysicians',JournalofDigitalImaging,vol.
23,no.1,pp.7380.
Eckhardt, A, Laumer, S & Weitzel, T 2009, 'Who Influences Whom Analyzing Workplace
ReferentsSocialInfluenceonItAdoptionandNonAdoption',JournalofInformation
Technology,vol.24,no.1,pp.1124.
Edwards, IR & Shipp, AI 2007, 'The Relationship between PersonEnvironment Fit and
Outcomes:AnIntegrative',Perspectivesonorganizationalfit,p.209.
Edwards, JR 1991, PersonJob Fit: A Conceptual Integration, Literature Review, and
MethodologicalCritique,JohnWiley&Sons.
1996,'AnExaminationofCompetingVersionsofthePersonEnvironmentFitApproach
toStress',Academyofmanagementjournal,vol.39,no.2,pp.292339.
Ehrhart,KH&Makransky,G2007,'TestingVocationalInterestsandPersonalityasPredictors
ofPersonVocationandPersonJobFit',JournalofCareerAssessment,vol.15,no.2,
pp.20626.
EscobarRodrguez, T & CarvajalTrujillo, E 2013, 'Online Drivers of Consumer Purchase of
WebsiteAirlineTickets',JournalofAirTransportManagement,vol.32,no.0,pp.58
64.
Farina,A&Miller,A1982,'TheStigmaofMentalDisorders',inMiller(ed.),IntheEyeofthe
Beholder,Praeger,NewYork,pp.30563.
Farrell, AM 2010, 'Insufficient Discriminant Validity: A Comment on Bove, Pervan, Beatty,
andShiu(2009)',JournalofBusinessResearch,vol.63,no.3,pp.3247.
Fazio,RH1986,'HowDoAttitudesGuideBehavior',Handbookofmotivationandcognition:
Foundationsofsocialbehavior,vol.1,pp.20443.
Fine,C2008,AMindofItsOwn:HowYourBrainDistortsandDeceives,WWNorton,New
York.
Fink,A2009,HowtoConductSurveys:AStepbyStepGuide,Sage,London.
Fischbein, SV 2011, 'The Use of Dreams in the Clinical Context: Convergencies and
Divergencies:AnInterdisciplinaryProposal1',InternationalJournalofPsychoanalysis,
vol.92,no.2,pp.33358.
Fishbein, M & Ajzen, I 1975, Belief, Attitude, Intention, and Behavior: An Introduction to
TheoryandResearch,AddisonWesleyPublishingCompany,Reading,Mass.
Forman,D2010,'SecondNatureandSpirit:HegelontheRoleofHabitintheAppearanceof
Perceptual Consciousness', The Southern Journal of Philosophy, vol. 48, no. 4, pp.
32552.
Fornell,C&Larcker,DF1981,'StructuralEquationModelswithUnobservableVariablesand
MeasurementError:AlgebraandStatistics',JournalofMarketingResearch,pp.382
8.
Foucault,M1973,TheOrderofThings,Tavistock,London.
Fujii, S & Kitamura, R 2003, 'What Does a OneMonth Free Bus Ticket Do to Habitual
Drivers?AnExperimentalAnalysisofHabitandAttitudeChange',Transportation,vol.
30,no.1,pp.81.

197

Gardner, B 2009, 'Modelling Motivation and Habit in Stable Travel Mode Contexts',
TransportationResearchPartF:TrafficPsychologyandBehaviour,vol.12,no.1,pp.
6876.
Gardner,B,Abraham,C,Lally,P&deBruijn,GJ2011,'TheHabitualUseoftheSelfReport
HabitIndex:AReply',AnnalsofBehavioralMedicine,pp.12.
Garrison, J 2003, 'Prophetic Epideictic Rhetoric: Poetic Education Beyond Good and Evil',
EducationalTheory,vol.53,no.2,pp.22141.
Garrison,JW2002,'HabitsasSocialToolsinContext',OTJR,vol.22,pp.117.
Gaskin, J 2012, The Stats Tools Package, Editi ^on edn,
http://statwiki.kolobkreations.com/index.php?title=Structural_Equation_Modeling&
oldid=4901.
Gefen,D2003,'TamorJustPlainHabit:ALookatExperiencedOnlineShoppers',Journalof
EndUserComputing,vol.15,no.3,pp.113.
Ghiselli, EE, Campbell, JP & Zedeck, S 1981, Measurement Theory for the Behavioral
Sciences,vol.8,WHFreemanSanFrancisco.
Gilligan,C1982,InaDifferentVoice:PsychologicalTheoryandWomen'sDevelopment,vol.
326,HarvardUniversityPress,MA.
Goldner, L & Levi, M 2014, 'Children's Family Drawings, Body Perceptions, and Eating
Attitudes:TheModeratingRoleofGender',TheArtsinPsychotherapy,vol.41,no.1,
pp.7988.
Goodhue, D & Thompson, RL 1995, 'TaskTechnology Fit and Individual Performance', MIS
Quarterly,vol.19,no.2,pp.21336.
Granger, DS 1998, 'Recovering the Everyday: John Dewey as Emersonian Pragmatist',
EducationalTheory,vol.48,no.3,pp.331.
Gravetter, FJ & Forzano, LAB 2010, Research Methods for the Behavioral Sciences
(Gravetter),CengageLearning,CA,USA.
Greenberg,J2009,'Introduction:DarwinandLiteraryStudies',TwentiethCenturyLiterature,
vol.55,no.4,pp.42344.
GreganPaxton, J, Hibbard, JD, Brunel, FF & Azar, P 2002, 'So That's What That Is:
ExaminingtheImpactofAnalogyonConsumers'KnowledgeDevelopmentforReally
NewProducts',Psychology&Marketing,vol.19,no.6,pp.53350.
Gruber, TR 1993, 'A Translation Approach to Portable Ontology Specifications', Knowledge
acquisition,vol.5,no.2,pp.199220.
Gu,S,Yan,J,Xiao,Z,Liu,N,Shen,D,Fan,W,Ji,L,Chen,Z&Chen,Y2010,'WhatAreDriving
Users to Click Ads? User Habit, Attitude, and Commercial Intention', ADKDD 2010:
The 4th International Workshop on Data Mining and Audience Intelligence for
Advertising,pp.19.
Guba,EG&Lincoln,YS1994,'CompetingParadigmsinQualitativeResearch',Handbookof
qualitativeresearch,vol.2,pp.16394.
Guerreiro, R, Pereira, CA & Frezatti, F 2006, 'Evaluating Management Accounting Change
According to the Institutional Theory Approach', Journal of Accounting &
OrganizationalChange,vol.2,no.3,pp.196.
Guinea,AOD&Markus,ML2009,'WhyBreaktheHabitofaLifetime?RethinkingtheRoles
of Intention, Habit, and Emotion in Continuing Information Technology Use', MIS
Quarterly,vol.33,no.3,pp.43344.

198

Gupta,B,Dasgupta,S&Gupta,A2008,'AdoptionofIctinaGovernmentOrganizationina
Developing Country: An Empirical Study', The Journal of Strategic Information
Systems,vol.17,no.2,pp.14054.
Hair, JF, Black, WC, Babin, BJ & Anderson, RE 2010, Multivariate Data Analysis: Global
Edition,7thedn,PearsonHigherEdUSA,NewJersey.
Hair, JF, Ringle, CM & Sarstedt, M 2011, 'PlsSem: Indeed a Silver Bullet', Journal of
MarketingTheoryandPractice,vol.19,no.2,pp.13951.
Haldimann,L,Walter,C&Brenzikofer,T2014,SwissMadeSoftwaretheBook,buch&netz.
Han,Y&Farn,C2013,'AStudyontheEffectsofEmpowermentandHabitonContinuance
Usage of Pervasive Business Intelligence Systems', paper presented to System
Sciences(HICSS),201346thHawaiiInternationalConferenceon.
Hanson, C, West, J, Neiger, B, Thackeray, R, Barnes, M & McIntyre, E 2011, 'Use and
Acceptance of Social Media among Health Educators', American Journal of Health
Education,vol.42,no.4,pp.197204.
Heckathorn,DD2002,'RespondentDrivenSamplingIi:DerivingValidPopulationEstimates
fromChainReferralSamplesofHiddenPopulations',SocialProblems,vol.49,no.1,
pp.1134.
Hedoin, C 2009, 'Weber and Veblen on the Rationalization Process', Journal of Economic
Issues,vol.43,no.1,pp.16787.
Hodgson,GM1997,'TheUbiquityofHabitsandRules',CambridgeJournalofEconomics,vol.
21,no.6,pp.66384.
Hodgson, GM 2007, 'Rationality Versus ProgramBased Behavior', Behavioral and Brain
Sciences,vol.30,pp.161.
2009,'OntheInstitutionalFoundationsofLaw:TheInsufficiencyofCustomandPrivate
Ordering',JournalofEconomicIssues,vol.43,no.1,pp.14366.
Hodgson, GM 2010, 'Choice, Habit and Evolution', Journal of Evolutionary Economics, vol.
20,no.1,pp.118.
Hodgson, GM & Thorbjrn, K 2006, 'Dismantling Lamarckism: Why Descriptions of Socio
Economic Evolution as Lamarckian Are Misleading', Journal of Evolutionary
Economics,vol.16,no.4,pp.34366.
Hofstede, G 1983, 'Dimensions of National Cultures in Fifty Countries and Three Regions',
Expiscationsincrossculturalpsychology,pp.33555.
Honkanen, P, Olsen, SO & Verplanken, B 2005, 'Intention to Consume Seafoodthe
ImportanceofHabit',Appetite,vol.45,no.2,pp.1618.
Horverak,JG,Bye,HH,Sandal,GM&Pallesen,S2013,'ManagersEvaluationsofImmigrant
Job Applicants the Influence of Acculturation Strategy on Perceived Person
Organization Fit (Po Fit) and Hiring Outcome', Journal of CrossCultural Psychology,
vol.44,no.1,pp.4660.
Hsu, CL, Tseng, KC & Chuang, YH 2011, 'Predictors of Future Use of Telehomecare Health
Services by MiddleAged People in Taiwan', Social Behavior and Personality: an
internationaljournal,vol.39,no.9,pp.125161.
Huang, TCK, Wu, IL & Chou, CC 2013, 'Investigating Use Continuance of Data Mining
Tools',InternationalJournalofInformationManagement,vol.33,no.5,pp.791801.
Hull,CL1943,PrinciplesofBehavior:AnIntroductiontoBehaviorTheory,AppletonCentury
Crofts.
Hume,D1984,ATreatiseofHumanNature,Penguin,Harmondsword.
199

Hung, YH, Wang, YS & Chou, SCT '63. User Acceptance of EGovernment Services', PACIS
2007Proceedings,97,pp.110.
Husserl, E 1970, The Crisis of the European Sciences and Transcendental Phenomenology,
NorthwestUniveristyPress,Evanston,IL.
HyeHyun,Y2013,'TheEffectsofOrganizationalServiceOrientationonPersonOrganization
FitandTurnoverIntent',ServiceIndustriesJournal,vol.33,no.1,pp.729.
Iacobucci, D 2010, 'Structural Equations Modeling: Fit Indices, Sample Size, and Advanced
Topics',JournalofConsumerPsychology,vol.20,no.1,pp.908.
Iijima, M, Arisaka, O, Minamoto, F & Arai, Y 2001, 'Sex Differences in Children's Free
Drawings: A Study on Girls with Congenital Adrenal Hyperplasia', Hormones and
Behavior,vol.40,no.2,pp.99104.
Isabelle, B & Sandrine, OH 2009, 'Towards an Understanding of Knowledge Management
SystemsUtautRevisited',AMCIS2009Proceedings,Paper445,pp.113.
James,W1890,ThePrinciplesofPsychology,VolI.
James, W, Burkhardt, F, Fredson , B & Skrupskelis, IK 1890, The Principles of Psychology,
HarvardUniversityPress.
Jansson, C, Wullschleger, SD, Kalluri, UC & Tuskan, GA 2010, 'Phytosequestration: Carbon
Biosequestration by Plants and the Prospects of Genetic Engineering', Bioscience,
vol.60,no.9,pp.68596.
Jiang, JJ, Muhanna, WA & Klein, G 2000, 'User Resistance and Strategies for Promoting
AcceptanceacrossSystemTypes',Information&Management,vol.37,no.1,pp.25
36.
Jreskog, KG & Srbom, D 1996, Lisrel 8: User's Reference Guide, 2nd edn, Scientific
SoftwareInternational,Chicago,IL.
Judge, TA & Cable, D 1997, 'Interviewers' Perceptions of PersonOrganization Fit and
OrganizationalSelectionDecisions',JournalofAppliedPsychology,vol.82,pp.546
61.
Junglas,I,Abraham,C&Ives,B2009,'MobileTechnologyattheFrontlinesofPatientCare:
Understanding Fit and Human Drives in Utilization Decisions and Performance',
DecisionSupportSystems,vol.46,no.3,pp.63447.
Kail, RV & Cavanaugh, JC 2011, Human Development: A LifeSpan View: A LifeSpan View,
6thedn,CengageLearning,CA,USA.
Kang,YS,Min,J,Kim,J&Lee,H2013,'RolesofAlternativeandSelfOrientedPerspectivesin
the Context of the Continued Use of Social Network Sites', International Journal of
InformationManagement,vol.33,no.3,pp.496511.
Kant,I2007,CritiqueofPureReason,Penguin,Harmondsworth.
Karahanna, E, Agarwal, R & Angst, CM 2006, 'Reconceptualizing Compatibility Beliefs in
TechnologyAcceptanceResearch',MISQuarterly,pp.781804.
Karahanna, E, Straub, DW & Chervany, NL 1999, 'Information Technology Adoption across
Time:ACrossSectionalComparisonofPreAdoptionandPostAdoptionBeliefs',MIS
Quarterly,pp.183213.
Kemp,S1998,'MedievalTheoriesofMentalRepresentation',HistoryofPsychology,vol.1,
no.4,pp.27588.
Kijsanayotin, B, Pannarunothai, S & Speedie, SM 2009, 'Factors Influencing Health
InformationTechnologyAdoptioninThailand'sCommunityHealthCenters:Applying

200

 the Utaut Model', International Journal of Medical Informatics, vol. 78, no. 6, pp.
40416.
Kim, SS & Malhotra, NK 2005, 'A Longitudinal Model of Continued Is Use: An Integrative
View of Four Mechanisms Underlying Postadoption Phenomena', Management
Science,vol.51,no.5,pp.74155.
Kim,TY&Kim,M2013,'LeadersMoralCompetenceandEmployeeOutcomes:TheEffects
of Psychological Empowerment and PersonSupervisor Fit', Journal of business
ethics,vol.112,no.1,pp.15566.
Klckner, CA 2013, 'A Comprehensive Model of the Psychology of Environmental
BehaviouraMetaAnalysis',GlobalEnvironmentalChange,vol.23,no.5,pp.1028
38.
Klckner, CA & Matthies, E 2004, 'How Habits Interfere with NormDirected Behaviour: A
Normative DecisionMaking Model for Travel Mode Choice', Journal of
EnvironmentalPsychology,vol.24,no.3,pp.31927.
Klckner, CA, Matthies, E & Hunecke, M 2003, 'Problems of Operationalizing Habits and
Integrating Habits in Normative DecisionMaking Models', Journal of Applied Social
Psychology,vol.33,no.2,pp.396417.
Kremers, SPJ, van der Horst, K & Brug, J 2007, 'Adolescent ScreenViewing Behaviour Is
Associated with Consumption of SugarSweetened Beverages: The Role of Habit
StrengthandPerceivedParentalNorms',Appetite,vol.48,no.3,pp.34550.
Krippendorff,K2012,ContentAnalysis:AnIntroductiontoItsMethodology,Sage.
KristofBrown,AL&Billsberry,J2012,'FitfortheFuture',OrganizationalFit:KeyIssuesand
NewDirections,pp.118.
KristofBrown,AL,Zimmerman,RD&Johnson,EC2005,'ConsequencesofIndividuals'Fitat
Work: A MetaAnalysis of PersonJob, PersonOrganization, PersonGroup, and
PersonSupervisorFit',Personnelpsychology,vol.58,no.2,pp.281342.
Kristof, AL 1996, 'PersonOrganization Fit: An Integrative Review of Its Conceptualizations,
Measurement,andImplications',Personnelpsychology,vol.49,no.1,pp.149.
Krugman, HE 1966, 'The Measurement of Advertising Involvement', Public Opinion
Quarterly,vol.30,no.4,pp.58396.
Ku, Th 2007, 'Psychoanalytic Semiotics and the Interpretation of Dream Paintings: An
ExamplefromSalvadorDali',TheAmericanJournalofSemiotics,vol.23,no.14,pp.
30336,78.
Lally, P, van Jaarsveld, CHM, Potts, HWW & Wardle, J 2010, 'How Are Habits Formed:
ModellingHabitFormationintheRealWorld',EuropeanJournalofSocialPsychology,
vol.40,no.6,pp.9981009.
Landis, D, Triandis, HC & Adamopoulos, J 1978, 'Habit and Behavioral Intentions as
Predictors of Social Behavior', The Journal of Social Psychology, vol. 106, no. 2, pp.
22737.
Lankton, NK, Wilson, EV & Mao, E 2010, 'Antecedents and Determinants of Information
TechnologyHabit',Information&Management,vol.47,no.56,pp.3007.
Lawrence, AD, Evans, AH & Lees, AJ 2003, 'Compulsive Use of Dopamine Replacement
TherapyinParkinson'sDisease:RewardSystemsGoneAwry?',TheLancetNeurology,
vol.2,no.10,pp.595604.
Ledgerwood, J & White, V 2006, Transforming Microfinance Institutions: Providing Full
FinancialServicestothePoor,WorldBankPublications.
201

Lee, CL, Yen, DC, Peng, KC & Wu, HC 2010, 'The Influence of Change Agents' Behavioral
IntentionontheUsageoftheActivityBasedCosting/ManagementSystemandFirm
Performance: The Perspective of Unified Theory of Acceptance and Use of
Technology',AdvancesinAccounting,vol.26,no.2,pp.31424.
Lee,K,Lee,S&Kim,J2005,'AnalysisofMobileCommercePerformancebyUsingtheTask
TechnologyFit',MobileInformationSystems,pp.13553.
Leech, NL, Barrett, KC & Morgan, GA 2011, Ibm Spss for Intermediate Statistics: Use and
Interpretation,4thedn,RoutledgeAcademic,NewYork,NY.
Lefebvre, M 2007, 'Peirce's Esthetics: A Taste for Signs in Art', Charles S. Peirce Society.
TransactionsoftheCharlesS.PeirceSociety,vol.43,no.2,pp.31944.
Legris, P, Ingham, J & Collerette, P 2003, 'Why Do People Use Information Technology? A
CriticalReviewoftheTechnologyAcceptanceModel',Information&Management,
vol.40,no.3,pp.191204.
Li, J & Kishore, R 2006, 'How Robust Is the Utaut Instrument?: A Multigroup Invariance
Analysis in the Context of Acceptance and Use of Online Community Weblog
Systems'.
Liang,H,Xue,YL,Ke,W&Wei,KK2010,'UnderstandingtheInfluenceofTeamClimateonIt
Use',JournaloftheAssociationforInformationSystems,vol.11,no.8,pp.41432.
Liao,C,Palvia,P&Lin,HN2006,'TheRolesofHabitandWebSiteQualityinECommerce',
InternationalJournalofInformationManagement,vol.26,no.6,pp.46983.
Liao, C, PuiLai, T, ChuangChun, L, PuYuan, K & ShuHui, C 2011, 'Factors Influencing the
IntendedUseofWebPortals',OnlineInformationReview,vol.35,no.2,pp.23754.
Liberman,N&Frster,J2008,'Expectancy,ValueandPsychologicalDistance:ANewLookat
GoalGradients',SocialCognition,vol.26,no.5,pp.51533.
Likert,R1932,'ATechniquefortheMeasurementofAttitudes',Archivesofpsychology,vol.
140,no.5.
Limayem, M, Cheung, C & Chan, G 2003, 'Explaining Information Systems Adoption and
PostAdoption:TowardanIntegrativeModel',ICIS2003Proceedings,59,pp.72031.
Limayem, M & Cheung, CM 2008, 'Understanding Information Systems Continuance: The
CaseofInternetBasedLearningTechnologies',Information&Management,vol.45,
no.4,pp.22732.
Limayem, M & Hirt, SG 2003, 'Force of Habit and Information Systems Usage: Theory and
Initial Validation', Journal of the Association for Information Systems, vol. 4, no. 1,
pp.6597.
Limayem,M,Hirt,SG&Cheung,CMK2003,'HabitintheContextofIsContinuance:Theory
ExtensionandScaleDevelopment',ECIS2003Proceedings,Paper90,pp.118.
2007, 'How Habit Limits the Predictive Power of Intention: The Case of Information
SystemsContinuance',MISQuarterly,vol.31,no.4,pp.70537.
Lin, J, Chan, H & Jin, Y 2004, 'Instant Messaging Acceptance and Use among College
Students',PACIS2004Proceedings,pp.15.
Lin, TC & Huang, CC 2009, 'Understanding the Determinants of Ekr Usage from Social,
TechnologicalandPersonalPerspectives',JournalofInformationScience,vol.35,no.
2,pp.165.
Lindbladh,E&Lyttkens,CH2002,'HabitVersusChoice:TheProcessofDecisionMakingin
HealthRelatedBehaviour',Socialscience&medicine,vol.55,no.3,pp.45165.

202

Lloyd, BB 1972, Perception and Cognition: A CrossCultural Perspective, Penguin

Harmondsworth.
Loibl,C,Kraybill,DS&DeMay,SW2011,'AccountingfortheRoleofHabitinRegularSaving',
JournalofEconomicPsychology,vol.32,no.4,pp.58192.
Louis, MR & Sutton, RI 1991, 'Switching Cognitive Gears: From Habits of Mind to Active
Thinking',HumanRelations,vol.44,no.1,pp.5576.
Lu, J, Yao, JE & Yu, CS 2005, 'Personal Innovativeness, Social Influences and Adoption of
Wireless Internet Services Via Mobile Technology', The Journal of Strategic
InformationSystems,vol.14,no.3,pp.24568.
Lu,J,Yu,CS&Liu,C2009,'MobileDataServiceDemographicsinUrbanChina',Journalof
ComputerInformationSystems,vol.50,no.2,pp.11726.
Luebben,AJ&Royeen,CB2007,'TowardVerstehen:AnEtymologicalandHistoricalWaveof
theTerms'Habit,''Routine,''Occupation,'and'Participation'',OTJR,vol.27,pp.867.
MacMullan, TA 2005, 'Is There a White Gift?: A Pragmatist Response to the Problem of
Whiteness', Charles S. Peirce Society. Transactions of the Charles S. Peirce Society,
vol.41,no.4,pp.796817.
Mair,J&Berginseers,S2010,'TheEffectofInterventionsontheEnvironmentalBehaviour
ofAustralianMotelGuests',TourismandHospitalityResearch,vol.10,no.4,pp.255
68.
Marcus, B & Wagner, U 2013, 'What Do You Want to Be? CriterionRelated Validity of
AttainedVocationalAspirationsVersusInventoriedPersonVocationFit',Journalof
BusinessandPsychology,pp.112.
Marchal, K 2010, 'Not Irrational but Habitual: The Importance Of', Ecological Economics,
vol.69,no.5,pp.110414.
Marescaux, J & Diana, M 2014, 'Robotics and Remote Surgery: Next Step', in Robotics in
GeneralSurgery,Springer,pp.47984.
Margolis, H 1990, Patterns, Thinking, and Cognition: A Theory of Judgment, University of
ChicagoPress,Chicago.
Marsh,HW,Balla,JR&McDonald,RP1988,'GoodnessofFitIndexesinConfirmatoryFactor
Analysis:TheEffectofSampleSize',PsychologicalBulletin,vol.103,no.3,pp.391
410.
Martini, MG, Hewage, CT & Nasralla, MM 2014, '3d Robotic Surgery and Training at a
Distance',in3dFutureInternetMedia,Springer,pp.25774.
Martins,C,Oliveira,T&Popovi,A2014,'UnderstandingtheInternetBankingAdoption:A
UnifiedTheoryofAcceptanceandUseofTechnologyandPerceivedRiskApplication',
InternationalJournalofInformationManagement,vol.34,no.1,pp.113.
Maruping, LM & Agarwal, R 2004, 'Managing Team Interpersonal Processes through
Technology: A TaskTechnology Fit Perspective* 1', Journal of Applied Psychology,
vol.89,no.6,pp.97590.
Mauss, M 1979, 'Techniques of the Body', in B Brewster (ed.), Sociology and Psychology
Routledge,London.
McGill,T,Klobas,J&Renzi,S2011,'LmsUseandInstructorPerformance:TheRoleofTask
TechnologyFit',InternationalJournalonELearning,vol.10,no.1,pp.4362.
McGuinness, TMPCF & McElroy, E 2010, '7 Habits of Highly Effective Psychiatric Nurse
Authors',JournalofPsychosocialNursing&MentalHealthServices,vol.48,no.1,pp.
338.
203

McLeod, A, Pippin, S & Catania, V 2009, 'Using Technology Acceptance Theory to Model
IndividualDifferencesinTaxSoftwareUse',AMCIS2009Proceedings,Paper811,pp.
111.
McLeod, JM & Chaffee, SH 1972, 'The Construction of Social Reality', The social influence
processes,pp.5099.
Melcalfe, J & Mischel, W 1999, 'A Hot/CoolSystem Analysis of Delay of Gratification:
DynamicsofWillpower',PsychologicalReview,no.106,pp.319.
UDoC National Institute of Science and Technology 2011, The Nist Definition of Cloud
Computing,Mell,P&Grance,T,vol.NISTSpecialPublication800145.
MerleauPonty,M1965,TheStructureofBehavior,Methuen,London.
Meyer, B & Sugiyama, K 2007, 'The Concept of Knowledge in Km: A Dimensional Model',
JournalofKnowledgeManagement,vol.11,no.1,pp.17.
MeyersLevy, J & Tybout, AM 1989, 'Schema Congruity as a Basis for Product Evaluation',
Journalofconsumerresearch,pp.3954.
Miller, HT 2010, 'The Oxford Handbook of Sociology and Organizational Studies: Classical
Foundations',AdministrativeTheory&Praxis,vol.32,no.3,pp.4738.
Milne, LC & Greenway, P 1999, 'Color in Childrens Drawings: The Influence of Age and
Gender',TheArtsinPsychotherapy,vol.26,no.4,pp.2613.
Mittal,B1988,'AchievingHigherSeatBeltUsage:TheRoleofHabitinBridgingtheAttitude
BehaviorGap1',JournalofAppliedSocialPsychology,vol.18,no.12,pp.9931016.
Montano, DE & Taplin, SH 1991, 'A Test of an Expanded Theory of Reasoned Action to
Predict Mammography Participation', Social science & medicine, vol. 32, no. 6, pp.
73341.
Moore, E, Richter, BA, Patton, CK & Lear, SA 2006, 'Mapping Stairwell Accessibility in
Vancouver'sDowntownCore',CanadianJournalofPublicHealth,vol.97,no.2,pp.
11820.
Moore,GC&Benbasat,I1991,'DevelopmentofanInstrumenttoMeasurethePerceptions
of Adopting an Information Technology Innovation', INFORMATION SYSTEMS
RESEARCH,vol.2,no.3,pp.192222.
Moore,P,Evans,C&Tadros,G2013,'TowardsIndependentAssistedLiving:TheApplication
of Intelligent ContextAware Systems', paper presented to Advanced Information
Networking and Applications Workshops (WAINA), 2013 27th International
Conferenceon.
Morris, MG & Venkatesh, V 2010, 'Job Characteristics and Job Satisfaction: Understanding
the Role of Enterprise Resource Planning System Implementation', MIS Quarterly,
vol.34,no.1,pp.14361.
Moses,O2009,'GertrudeStein'sLivelyHabits',TwentiethCenturyLiterature,vol.55,no.4,
pp.44584.
Muchinsky, PM & Monahan, CJ 1987, 'What Is PersonEnvironment Congruence?
Supplementary Versus Complementary Models of Fit', Journal of Vocational
Behavior,vol.31,no.3,pp.26877.
Mugge, R & Dahl, DW 2013, 'Seeking the Ideal Level of Design Newness: Consumer
ResponsetoRadicalandIncrementalProductDesign',JournalofProductInnovation
Management,vol.30,no.S1,pp.3447.
Mullany, MJ, Tan, FB & Gallupe, RB 2007, 'The Impact of AnalystUser Cognitive Style
DifferencesonUserSatisfaction'.
204

Muthn, B 1984, 'A General Structural Equation Model with Dichotomous, Ordered
Categorical, and Continuous Latent Variable Indicators', Psychometrika, vol. 49, no.
1,pp.11532.
Nahm,AY,SolsGalvn,LE,Rao,SS&RaguNathan,T2002,'TheQSortMethod:Assessing
Reliability and Construct Validity of Questionnaire Items at a PreTesting Stage',
JournalofModernAppliedStatisticalMethods,vol.1,no.1,pp.11425.
Nakamura, K 2009, 'The Significance of Dewey's Aesthetics in Art Education in the Age of
Globalization',EducationalTheory,vol.59,no.4,pp.42740.
Nslindhylispangar, A, Sihvonen, M, Sarna, S, Vanhanen, H & Kekki, P 2008, 'Has Brief
CounsellinganImpactontheRiskFactorsforCardiovascularDiseasein40YearOld
Men. A ThreeYear Followup Study', Central European Journal of Medicine, vol. 3,
no.3,pp.3007.
Neufeld, DJ, Dong, L & Higgins, C 2007, 'Charismatic Leadership and User Acceptance of
InformationTechnology',EuropeanJournalofInformationSystems,vol.16,no.4,pp.
494510.
Neuman,WL1994,SocialResearchMethods:QualitativeandQuantitativeApproaches,Editi
^onedn,AllynandBacon(Boston),0205145485.
Neuman, WL 2010, Social Research Methods: Qualitative and Quantitative Approaches:
InternationalEdition,7thedn,Pearson,USA.
Newton,CJ&Jimmieson,NL2008,'RoleStressors,ParticipativeControl,andSubjectiveFit
with Organisational Values: Main and Moderating Effects on Employee Outcomes',
JournalofManagementandOrganization,vol.14,no.1,pp.209,312,49.
Ney, B 2013, 'Unraveling the Adoption of Mcrm Smartphone Applications among Dutch
Retailers'.
Niessen,C,Swarowsky,C&Leiz,M2010,'AgeandAdaptationtoChangesintheWorkplace',
JournalofManagerialPsychology,vol.25,no.4,pp.35683.
Nikou,S&Bouwman,H2013,'TheDiffusionofMobileSocialNetworkServiceinChina:The
Role of Habit and Social Influence', paper presented to System Sciences (HICSS),
201346thHawaiiInternationalConferenceon.
Nistor, N, Baltes, B, Dasclu, M, Mihil, D, Smeaton, G & TruanMatu, 2013,
'Participation in Virtual Academic Communities of Practice under the Influence of
Technology Acceptance and Community Factors. A Learning Analytics Application',
ComputersinHumanBehavior,no.0.
Norman,P2011,'TheTheoryofPlannedBehaviorandBingeDrinkingamongUndergraduate
Students:AssessingtheImpactofHabitStrength',AddictiveBehaviors,vol.36,no.5,
pp.5027.
Norros, L & Salo, L 2009, 'Design of Joint Systems: A Theoretical Challenge for Cognitive
SystemsEngineering',Cognition,Technology&Work,vol.11,no.1,pp.4356.
Nunnally,JC&Bernstein,IH1994,PsychometricTheory,McGrawHill,NewYork.
Oconnor, BP 2000, 'Spss and Sas Programs for Determining the Number of Components
Using Parallel Analysis and Velicers Map Test', Behavior research methods,
instruments,&computers,vol.32,no.3,pp.396402.
Orbell, S, Blair, C, Sherlock, K & Conner, M 2001, 'The Theory of Planned Behavior and
Ecstasy Use: Roles for Habit and Perceived Control over Taking Versus Obtaining
Substances',JournalofAppliedSocialPsychology,vol.31,no.1,pp.3147.
Osgood,CE1957,TheMeasurementofMeaning,UniversityofIllinoispress,IL.
205

Otten, TJ 1999, 'The Spoils of Poynton and the Properties of Touch', American Literature,
vol.71,no.2,pp.26390.
Ouellette,JA&Wood,W1998,'HabitandIntentioninEverydayLife:TheMultipleProcesses
by Which Past Behavior Predicts Future Behavior', Psychological Bulletin, vol. 124,
no.1,pp.5474.
Pahnila, S, Siponen, M & Zheng, X 2011, 'Integrating Habit into Utaut: The Chinese Ebay
Case',PacificAsiaJournaloftheAssociationforInformationSystems,vol.3,Article2,
no.2,pp.130.
Pahnila,S&Warsta,J2010,'OnlineShoppingViewedfromaHabitandValuePerspective',
Behaviour&InformationTechnology,vol.29,no.6,pp.62132.
Pan, G, Hackney, R & Pan, SL 2008, 'Information Systems Implementation Failure: Insights
from Prism', International Journal of Information Management, vol. 28, no. 4, pp.
25969.
Pasticci, S 2009, 'Using History, to Make History: Goffredo Petrassi and the Heritage of
MusicalTradition',StudiMusicali,vol.38,no.2,pp.44585.
Pavlov,I1911,ConditionedReflexes,OxfordUniversityPress,Oxford.
Peng,C,Xueming,S,Hongyong,Y&Dengsheng,L2011,'AssessingTemporalandWeather
Influences on Property Crime in Beijing, China', Crime, Law and Social Change, vol.
55,no.1,pp.113.
Peterson,RA2000,ConstructingEffectiveQuestionnaires,Sage,USA.
Petillo, LM 2010, 'The Theological Problem of Grace and Experience: A Lonerganian
Perspective',TheologicalStudies,vol.71,no.3,pp.586608.
Petty, RE, Fazio, RH & Briol, P 2012, Attitudes: Insights from the New Implicit Measures,
PsychologyPress.
Polites, GL 2005, 'Counterintentional Habits as an Inhibitor of Technology Acceptance',
Proceedings of the 2005 Southern Association for Information Systems Conference,
pp.26471.
2009, 'The Duality of Habit in Information Technology Acceptance', University of
Georgia.
Polkowski,L2013,RoughSets:MathematicalFoundations,SpringerScience&Business,NY.
Poole,JL2000,'HabitsinWomenwithChronicDisesae:APilotStudy',OTJR,vol.20,pp.112
8.
Powell, A, Williams, CK, Bock, DB, Doellman, T & Allen, J 2012, 'EVoting Intent: A
Comparison of Young and Elderly Voters', Government Information Quarterly, vol.
29,no.3,pp.36172.
Powers,TL&Loyka,JJ2007,'Market,Industry,andCompanyInfluencesonGlobalProduct
Standardization',InternationalMarketingReview,vol.24,no.6,pp.67894.
Punch, KF 2005, Introduction to Social Research: Quantitative and Qualitative Approaches,
Sage.
PwC2013,PwcGlobal100SoftwareLeaders:ConvergingForcesAreBuildingThatCouldRe
ShapetheEntireIndustry,viewed2013/12/15<www.pwc.com/globalsoftware100>.
Rachlin,H2002,'AltruismandSelfishness',BehavioralandBrainSciences,vol.25,no.2,pp.
23996.
Ralston,SJ2011,'AMorePracticalPedagogicalIdeal:SearchingforaCriterionofDeweyan
Growth',EducationalTheory,vol.61,no.3,pp.35164.

206

Raman, A & Don, Y 2013, 'Preservice Teachers Acceptance of Learning Management

Software:AnApplicationoftheUTAUT2Model',InternationalEducationStudies,vol.
6,no.7,pp.15764.
Raposa, ML 2006, 'Phenomenology as Phaneroscopy: Theology in a New Key', American
JournalofTheology&Philosophy,vol.27,no.1,pp.8599,115.
Rein,JA1997,'VariablesandVennDiagrams',TheUniversityofArizona.
Resick,CJ,Baltes,BB&WalkerShantz,C2007,'PersonOrganizationFitandWorkRelated
Attitudes and Decisions: Examining Interactive Effects with Job Fit and
Conscientiousness',JournalofAppliedPsychology,vol.92,no.5,pp.144655.
Reunis, MRB, Santema, SC & Harink, JHA 2006, 'Increasing EOrdering Adoption: A Case
Study',JournalofPurchasingandSupplyManagement,vol.12,no.6,pp.32231.
Reynolds,L1981,'FoundationsofanInstitutionalTheoryofRegulation',JournalofEconomic
Issues(pre1986),vol.15,no.000003,pp.641.
Rhodes, RE & de Bruijn, GJ 2010, 'Automatic and Motivational Correlates of Physical
Activity: Does Intensity Moderate the Relationship?', Behavioral Medicine, vol. 36,
no.2,pp.4452.
Robinson, JP, Shaver, PR & Wrightsman, LS 1991, 'Criteria for Scale Selection and
Evaluation',Measuresofpersonalityandsocialpsychologicalattitudes,vol.1,pp.1
16.
Rogers,EM2003,DiffusionofInnovations,5thedn,FreePress,NewYork,NY.
Rogers,JC2000,'Habits:DoWePracticeWhatWePreach?',OTJR,vol.20,pp.11922.
Ronald Lee, Z 1998, 'The Four Domains of Moral Education: The Contributions of Dewey,
AlexanderandGolemantoaComprehensiveTaxonomy',JournalofMoralEducation,
vol.27,no.1,pp.1933.
Ronis,DL,Yates,JF&Kirscht,JP1989,'Attitudes,Decisions,andHabitsasDeterminantsof
RepeatedBehavior',Attitudestructureandfunction,pp.21339.
Rozin, P 2001, 'Social Psychology and Science: Some Lessons from Solomon Asch',
PersonalityandSocialPsychologyReview,vol.5,no.1,pp.214.
Saba,A&DiNatale,R1998a,'Attitudes,IntentionandHabit:TheirRoleinPredictingActual
ConsumptionofFatsandOils',JournalofHumanNutritionandDietetics,vol.11,no.
1,pp.2132.
1998b,'AStudyontheMediatingRoleofIntentionintheImpactofHabitandAttitude
onMeatConsumption',FoodQualityandPreference,vol.10,no.1,pp.6977.
Saba, A, Moneta, E, Nardo, N & Sinesio, F 1998, 'Attitudes, Habit, Sensory and Liking
Expectation as Determinants of the Consumption of Milk', Food Quality and
Preference,vol.9,no.12,pp.3141.
Saba, A, Vassallo, M & Turrini, A 2000, 'The Role of Attitudes, Intentions and Habit in
PredictingActualConsumptionofFatContainingFoodsinItaly',Europeanjournalof
clinicalnutrition,vol.54,no.7,pp.5405.
Sahu,GP&Gupta,M2007,'Users'AcceptanceofEGovernment:AStudyofIndianCentral
Excise',InternationalJournalofElectronicGovernmentResearch(IJEGR),vol.3,no.3,
pp.121.
Saks,AM&Ashforth,BE2002,'IsJobSearchRelatedtoEmploymentQuality?ItAllDepends
ontheFit',JournalofAppliedPsychology,vol.87,no.4,pp.64654.

207

Satyaprasad, S 2009, 'An Unusual Type of Sucking Habit in a Patient with Cleft Lip and
Palate',JournaloftheIndianSocietyofPedodonticsandPreventiveDentistry,vol.27,
no.4,pp.2602.
Saunders, MN, Saunders, M, Lewis, P & Thornhill, A 2011, Research Methods for Business
Students,5/E,PearsonEducation.
Schfer,M,Jaegererben,M&Bamberg,S2012,'LifeEventsasWindowsofOpportunityfor
ChangingTowardsSustainableConsumptionPatterns?',JournalofConsumerPolicy,
vol.35,no.1,pp.6584.
Schaper, LK & Pervan, GP 2007, 'An Investigation of Factors Affecting Technology
Acceptance and Use Decisions by Australian Allied Health Therapists', IEEE
Proceedingsofthe40thHawaiiInternationalConferenceonSystemSciences,pp.1
10.
Schaupp, LC, Carter, L & Hobbs, J 2010, 'EFile Adoption: A Study of Us Taxpayers'
Intentions',ComputersinHumanBehavior,vol.26,pp.636344.
Schmuck, P & Vlek, C 2003, 'Psychologists Can Do Much to Support Sustainable
Development',EuropeanPsychologist,vol.8,no.2,pp.6676.
Schreiber,JB,Nora,A,Stage,FK,Barlow,EA&King,J2006,'ReportingStructuralEquation
Modeling and Confirmatory Factor Analysis Results: A Review', The Journal of
EducationalResearch,vol.99,no.6,pp.32338.
Schultz,PW&Searleman,A2002,'RigidityofThoughtandBehavior:100YearsofResearch',
Genetic,Social,andGeneralPsychologyMonographs,vol.128,no.2,pp.165207.
Schwabe, L, Dickinson, A & Wolf, OT 2011, 'Stress, Habits, and Drug Addiction: A
Psychoneuroendocrinological Perspective', Experimental and Clinical
Psychopharmacology,vol.19,no.1,pp.5363.
Seger,CA1994,'ImplicitLearning',PsychologicalBulletin,vol.115,no.2,pp.16396.
Seton,PH2001,'OntheImportanceofGettingEven:AStudyoftheOriginsandIntentionof
Revenge',SmithCollegeStudiesinSocialWork,vol.72,no.1,pp.77.
Sharma, A 2005, Text Book of Sampling and Attributes, Discovery Publishing House, New
Delhi.
Sheeran,P,Aarts,H,Custers,R,Rivis,A&etal.2005,'TheGoalDependentAutomaticityof
DrinkingHabits',TheBritishJournalofSocialPsychology,vol.44,no.1,pp.4763.
Shields,PM2008,'RediscoveringtheTaproot:IsClassicalPragmatismtheRoutetoRenew
PublicAdministration?',PublicAdministrationReview,vol.68,no.2,pp.20521.
Simon, H 2001, 'On Simulating Simon: His Monomania, and Its Sources in Bounded
Rationality', Studies In History and Philosophy of Science Part A, vol. 32, no. 3, pp.
5015.
Simon,HA1974,'TheStructureofIllStructuredProblems',ArtificialIntelligence,vol.4,no.
3,pp.181201.
1997, Models of Bounded Rationality, Vol. 3: Emperically Grounded Economic Reason,
vol.3,MITPress,Cambridge,Mass.
Singer,M1981,'OntheSemioticsofIndianIdentity',TheAmericanJournalofSemiotics,vol.
1,no.1/2,pp.85126.
Sniehotta,FF&Presseau,J2012,'TheHabitualUseoftheSelfReportHabitIndex',Annalsof
BehavioralMedicine,vol.43,no.1,pp.13940.

208

Sok Foon, Y & Chan Yin Fah, B 2011, 'Internet Banking Adoption in Kuala Lumpur: An
ApplicationofUtautModel',InternationalJournalofBusinessandManagement,vol.
6,no.4,pp.1617.
Sonnentag,S&Jelden,S2009,'JobStressorsandthePursuitofSportActivities:ADayLevel
Perspective',JournalofOccupationalHealthPsychology,vol.14,no.2,pp.16581.
Srite,M&Karahanna,E2006,'TheRoleofEspousedNationalCulturalValuesinTechnology
Acceptance',MISQuarterly,vol.30,no.3,pp.679704.
Staats, AW1959, 'Verbal HabitFamilies, Concepts, and the Operant Conditioning of Word
Classes',PsychologicalReview,vol.68,no.3,pp.190204.
Stapel, J 1969, 'About 35 Years of Market Research in the Netherlands', Markonderzock
Kwartaalschrift,vol.2,pp.37.
Stengel, BS 2010, 'The Complex Case of Fear and Safe Space', Studies in Philosophy and
Education,vol.29,no.6,pp.52340.
Stephenson,W1935,'TechniqueofFactorAnalysis',Nature.
1988,ThePlayTheoryofMassCommunication,TransactionBooks.
Stopher, PR & Metcalf, HM 1996, Methods for Household Travel Surveys, vol. 236,
TransportationResearchBoard.
Strack, F & Deutsch, R 2004, 'Reflective and Impulsive Determinants of Social Behavior',
PersonalityandSocialPsychologyReview,vol.8,no.3,pp.22047.
Suksangiam, W & Chaiyasoonthorn, W 2013, 'The Adoption of Social Media by Thai
University Students Multiple Group Moderating Effects', Information Development,
p.0266666913502800.
Swartz,DL2002,'TheSociologyofHabit:ThePerspectiveofPierreBourdieu',OTJR,vol.22,
pp.619.
Sweet, S, Roome, N & Sweet, P 2003, 'Corporate Environmental Management and
SustainableEnterprise:TheInfluenceofInformationProcessingandDecisionStyles',
BusinessStrategyandtheEnvironment,vol.12,no.4,pp.265.
Sydnor,S,Sass,MSM,Adeola,M&Snuggs,T2014,'QualitativeAnalysisofMultidisciplinary
CollegeStudentsinanInternationalAlternativeBreakCourse',TheOnlineJournalof
QualityinHigherEducation,p.27.
Tabachnick, BG & Fidell, LS 2007, Using Multivariate Statistics, 5th edn, Pearson/Allyn &
Bacon,Boston.
Tabachnick,BG,Fidell,LS&Osterlind,SJ2001,UsingMultivariateStatistics.
Taylor,S&Todd,P1995,'AssessingItUsage:TheRoleofPriorExperience',MISQuarterly,
vol.19,no.4,pp.56170.
Teo,T,Lee,CB&Chai,CS2008,'UnderstandingPreServiceTeachers'ComputerAttitudes:
Applying and Extending the Technology Acceptance Model', Journal of computer
assistedlearning,vol.24,no.2,pp.12843.
Theuvsen, L 2004, 'Doing Better While Doing Good: Motivational Aspects of Payfor
Performance Effectiveness in Nonprofit Organizations', Voluntas, vol. 15, no. 2, pp.
11736.
Thgersen, J 2006, 'Understanding Repetitive Travel Mode Choices in a Stable Context: A
PanelStudyApproach',TransportationResearchPartA:PolicyandPractice,vol.40,
no.8,pp.62138.
Thgersen,J&Mller,B2008,'BreakingCarUseHabits:TheEffectivenessofaFreeOne
MonthTravelcard',Transportation,vol.35,no.3,pp.32945.
209

Thomas,DM&Watson,RT2002,'QSortingandMisResearch:APrimer',Communications
oftheAssociationforInformationSystems,vol.8,no.1,pp.14156.
Thompson,RL,Higgins,CA&Howell,JM1991,'PersonalComputing:TowardaConceptual
ModelofUtilization',MISQuarterly,vol.15,no.1,pp.12543.
Tobias, R 2009, 'Changing Behavior by Memory Aids: A Social Psychological Model of
Prospective Memory and Habit Development Tested with Dynamic Field Data',
PsychologicalReview,vol.116,no.2,pp.40838.
Torres Maldonado, PU, Khan, GF, Moon, J & Rho, JJ 2011, 'ELearning Motivation and
EducationalPortalAcceptanceinDevelopingCountries',OnlineInformationReview,
vol.35,no.1,pp.6685.
Towler, G & Shepherd, R 1992, 'Modification of Fishbein and Ajzen's Theory of Reasoned
ActiontoPredictChipConsumption',FoodQualityandPreference,vol.3,no.1,pp.
3745.
Trafimow, D 2000, 'Habit as Both a Direct Cause of Intention to Use a Condom and as a
Moderator of the AttitudeIntention and Subjective NormIntention Relations',
PsychologyandHealth,vol.15,no.3,pp.38393.
Trafimow,DA&Wyer,RS1993,'TheCognitiveRepresentationofMundaneSocialEvents',
JournalofPersonalityandSocialPsychology,vol.64,no.3,pp.36576.
Triandis,HC1977,InterpersonalBehavior,Brooks/ColePub.Co.,Monterey,Calif.
Triandis,HC1979,'Values,Attitudes,andInterpersonalBehavior',NebraskaSymposiumon
Motivation,vol.27,pp.195259.
Tseng, CM, Chang, HL & Woo, TH 2013, 'Modeling Motivation and Habit in Driving
BehaviorunderLifetimeDriver'sLicenseRevocation',AccidentAnalysis&Prevention,
vol.51,no.0,pp.2607.
Tuorila, H & Pangborn, R 1988, 'Behavioural Models in the Prediction of Consumption of
SelectedSweet,SaltyandFattyFoods',Foodacceptability,pp.26779.
Tyrvinen,P&Selin,J2011,'HowtoSellSaas:AModelforMainFactorsofMarketingand
SellingSoftwareasaService',inSoftwareBusiness,Springer,pp.216.
vanBerkel,J,Proper,KI,Boot,CRL,Bongers,PM&vanderBeek,AJ2011,'Mindful"Vitality
inPractice":AnInterventiontoImprovetheWorkEngagementandEnergyBalance
amongWorkers;theDevelopmentandDesignoftheRandomisedControlledTrial',
BMCPublicHealth,vol.11,no.1,pp.112.
van Empelen, P & Kok, G 2006, 'Condom Use in Steady and Casual Sexual Relationships:
Planning,PreparationandWillingnesstoTakeRisksamongAdolescents',Psychology
&Health,vol.21,no.2,pp.16581.
VanSickle,J1997,'UsingMeanSimilarityDendrogramstoEvaluateClassifications',Journal
ofAgricultural,Biological,andEnvironmentalStatistics,pp.37088.
van Vianen, AE, De Pater, IE & Van Dijk, F 2007, 'Work Value Fit and Turnover Intention:
SameSourceorDifferentSourceFit',JournalofManagerialPsychology,vol.22,no.
2,pp.188202.
Vandenberg, BR 2010, 'Evidence, Ontology, and Psychological Science: The Lesson of
Hypnosis',JournalofTheoreticalandPhilosophicalPsychology,vol.30,no.1,pp.51
65.
VanderStoep, SW & Johnston, DD 2009, Research Methods for Everyday Life: Blending
QualitativeandQuantitativeApproaches,JosseyBass,SanFrancisco,CA.

210

Veblen,T1898,'WhyIsEconomicsNotanEvolutionaryScience?',TheQuarterlyJournalof
Economics,pp.37397.
Velamuri, SR & Dew, N 2010, 'Evolutionary Processes, Moral Luck, and the Ethical
ResponsibilitiesoftheManager',Journalofbusinessethics,vol.91,no.1,pp.11326.
Venkatesh, V & Davis, FD 2000, 'A Theoretical Extension of the Technology Acceptance
Model:FourLongitudinalFieldStudies',ManagementScience,pp.186204.
Venkatesh,V,Davis,FD&Morris,MG2007a,'DeadorAlive?TheDevelopment,Trajectory
and Future of Technology Adoption Research', Journal of the Association for
InformationSystems,vol.8,no.4,pp.26786.
2007b, 'Dead or Alive? The Evolution, Trajectory, and Future of Technology Adoption
Research',JournaloftheAssociationforInformationSystems,vol.8,no.4,pp.267
86.
Venkatesh, V, Morris, MG, Davis, GB & Davis, FD 2003, 'User Acceptance of Information
Technology:TowardaUnifiedView',MISQuarterly,vol.27,no.3,pp.42578.
Venkatesh, V, Thong, J & Xu, X 2012, 'Consumer Acceptance and Use of Information
Technology: Extending the Unified Theory of Acceptance and Use of Technology',
ManagementInformationSystemsQuarterly,vol.36,no.1,pp.15778.
Venkatesh,V & Zhang,XJ 2010, 'Unified Theory of Acceptanceand Use of Technology: Us
Vs.China',JournalofGlobalInformationTechnologyManagement,vol.13,no.1,pp.
527.
Venkatraman, N & Camillus, JC 1984, 'Exploring the Concept of" Fit" in Strategic
Management',TheAcademyofManagementReview,vol.9,no.3,pp.51325.
Verplanken,B2006,'BeyondFrequency:HabitasMentalConstruct',BritishJournalofSocial
Psychology,vol.45,no.3,pp.63956.
Verplanken,B&Aarts,H1999,'Habit,Attitude,andPlannedBehaviour:IsHabitanEmpty
ConstructoranInterestingCaseofGoalDirectedAutomaticity?',Europeanreviewof
socialpsychology,vol.10,no.1,pp.10134.
Verplanken,B,Aarts,H,Advan,K&Moonen,A1998,'HabitVersusPlannedBehaviour:A
FieldExperiment',TheBritishJournalofSocialPsychology,vol.37,pp.11128.
Verplanken, B, Aarts, H, Knippenberg, A & Moonen, A 1998, 'Habit Versus Planned
Behaviour:AFieldExperiment',BritishJournalofSocialPsychology,vol.37,no.1,pp.
11128.
Verplanken,B,Aarts,H&VanKnippenberg,A1997,'Habit,InformationAcquisition,andthe
ProcessofMakingTravelModeChoices',EuropeanJournalofSocialPsychology,vol.
27,no.5,pp.53960.
Verplanken,B,Aarts,H.,vanKnippenberg,A.,&vanKnippenberg,C.1994,'AttitudeVersus
General Habit: Antecedents of Travel Mode Choice', Journal of Applied Social
Psychology,vol.24,pp.285300.
Verplanken, B & Faes, S 1999, 'Good Intentions, Bad Habits, and Effects of Forming
Implementation Intentions on Healthy Eating', European Journal of Social
Psychology,vol.29,no.56,pp.591604.
Verplanken, B, Friborg, O, Wang, CE, Trafimow, D & Woolf, K 2007, 'Mental Habits:
MetacognitiveReflectiononNegativeSelfThinking',JournalofPersonalityandSocial
Psychology,vol.92,no.3,pp.52641.
Verplanken, B & Melkevik, O 2008, 'Predicting Habit: The Case of Physical Exercise',
Psychologyofsportandexercise,vol.9,no.1,pp.1526.
211

Verplanken, B, Myrbakk, V & Rudi, E 2005, 'The Measurement of Habit', The routines of
decisionmaking,pp.23147.
Verplanken,B&Orbell,S2003,'ReflectionsonPastBehavior:ASelfReportIndexofHabit
Strength',JournalofAppliedSocialPsychology,vol.33,no.6,pp.131330.
Verplanken, B, Walker, I, Davis, A & Jurasek, M 2008, 'Context Change and Travel Mode
Choice: Combining the Habit Discontinuity and SelfActivation Hypotheses', Journal
ofEnvironmentalPsychology,vol.28,no.2,pp.1217.
Verplanken, B & Wood, W 2006, 'Interventions to Break and Create Consumer Habits',
JournalofPublicPolicy&Marketing,pp.90103.
Verquer,ML2002,'FittinginatWork:AComparisonoftheRelationshipsbetweenPerson
Organization Fit and PersonGroup Fit with Work Attitudes', Ph.D. thesis, Central
MichiganUniversity.
Wacquant,L2005,'Habitus',Internationalencyclopediaofeconomicsociology,pp.31519.
Wang, C, Liu, W, Tseng, M & Tsai, H 2010, 'A Study of Taiwanese College Teachers
AcceptanceofDistanceLearning',InternationalJournalofOrganizationalInnovation,
vol.3,no.2,pp.24360.
Wang, HY & Wang, SH 2010, 'User Acceptance of Mobile Internet Based on the Unified
Theory of Acceptance and Use of Technology: Investigating the Determinants and
Gender Differences', Social Behavior and Personality: an international journal, vol.
38,no.3,pp.41526.
Wang, YS & Shih, YW 2009, 'Why Do People Use Information Kiosks? A Validation of the
Unified Theory of Acceptance and Use of Technology', Government Information
Quarterly,vol.26,no.1,pp.15865.
Waterman,MS&Smith,TF1978,'OntheSimilarityofDendrograms',JournalofTheoretical
Biology,vol.73,no.4,pp.789800.
Watson,JB1998,Behaviorism,TransactionPublishers,NewBrunswick,N.J.
2008,Behavior:AnIntroductiontoComparativePsychology,WatsonPress,NewYork.
Weber,M2004,PuritanismandtheSpiritofCapitalism,andConfucianismandPuritanism
ComparedRoutledge,London.
Welty,C&Guarino,N2001,'SupportforOntologicalAnalysisofTaxonomicRelationships',
J.DataandKnowledgeEngineering,vol.39,no.1,pp.5174
West, SG, Finch, JF & Curran, PJ 1995, 'Structural Equation Models with Nonnormal
Variables:ProblemsandRemedies',inRHHoyle(ed.),StructuralEquationModeling:
Concepts, Issues, and Applications,SAGE Publications. Inc., Thousand Oaks, CA, pp.
5675.
Westland,JC&Clark,TH2001,GlobalElectronicCommerce:TheoryandCaseStudies(Mit
Press),UniversitiesPress.
Wheaton,B,Muthn,B,Alwin,DF&Summers,GF1977,'AssessingReliabilityandStability
inPanelModels',Sociologicalmethodology,vol.9,pp.84136.
Whisman,MA&McClelland,GH2005,'Designing,Testing,andInterpretingInteractionsand
ModeratorEffectsinFamilyResearch',JournalofFamilyPsychology,vol.19,no.1,
pp.11120.
White,TL&McBurney,DH2010,ResearchMethods,9thedn,Belmont,CA.
Wittenbraker,J,Gibbs,BL&Kahle,LR1983,'SeatBeltAttitudes,Habits,andBehaviors:An
AdaptiveAmendmenttotheFishbeinModel',JournalofAppliedSocialPsychology,
vol.13,no.5,pp.40621.
212

Wood, W & Neal, DT 2007, 'A New Look at Habits and the HabitGoal Interface',
PsychologicalReview,vol.114,no.4,pp.84363.
2009,'TheHabitualConsumer',JournalofConsumerPsychology,vol.19,no.4,pp.579
92.
Wood, W, Quinn, JM & Kashy, DA 2002, 'Habits in Everyday Life: Thought, Emotion, and
Action',JournalofPersonalityandSocialPsychology,vol.83,no.6,pp.128197.
Wood,W,Tam,L&Witt,MG2005,'ChangingCircumstances,DisruptingHabits',Journalof
PersonalityandSocialPsychology,vol.88,no.6,p.918.
Wozniak, RH 2009, 'Consciousness, Social Heredity, and Development: The Evolutionary
ThoughtofJamesMarkBaldwin',AmericanPsychologist,vol.64,no.2,pp.93101.
Wu,MC&Kuo,FY2008,'AnEmpiricalInvestigationofHabitualUsageandPastUsageon
Technology Acceptance Evaluations and Continuance Intention', ACM SIGMIS
Database,vol.39,no.4,pp.4873.
Yang, K 2010, 'Determinants of Us Consumer Mobile Shopping Services Adoption:
Implications for Designing Mobile Shopping Services', Journal of Consumer
Marketing,vol.27,no.3,pp.26270.
Yen,DC,Wu,CS,Cheng,FF&Huang,YW2010,'DeterminantsofUsers'IntentiontoAdopt
WirelessTechnology:AnEmpiricalStudybyIntegratingTtfwithTam',Computersin
HumanBehavior,vol.26,no.5,pp.90615.
Yeow, PHP & Loo, WH 2009, 'Acceptability of Atm and Transit Applications Embedded in
Multipurpose Smart Identity Card: An Exploratory Study in Malaysia', International
JournalofElectronicGovernmentResearch(IJEGR),vol.5,no.2,pp.3756.
Yoon,C2011,'TheoryofPlannedBehaviorandEthicsTheoryinDigitalPiracy:AnIntegrated
Model',Journalofbusinessethics,vol.100,no.3,pp.40517.
Yung, YF & Bentler, PM 1996, 'Bootstrapping Techniques in Analysis of Mean and
Covariance Structures', Advanced structural equation modeling: Issues and
techniques,pp.195226.
Zhang, P & Sun, H 2009, 'The Complexity of Different Types of Attitudes in Initial and
Continued Ict Use', Journal of the American Society for Information Science and
Technology,vol.60,no.10,pp.204863.
Zhu, W 1997, 'Making Bootstrap Statistical Inferences: A Tutorial', Research Quarterly for
ExerciseandSport,vol.68,no.1,pp.4455.
Zikmund,WG2003,BusinessResearchMethods,7thedn,CENAGELearning,Mason,OH.

213

 APPENDIX1KEYCHARACTERISTICSOFHABITEXTRACTEDFROMDEFINITIONS

Standardized
(Plurals and
No Reference Specific Discipline Grouped Discipline Definition Grouped as:
Regionalism s
elim inated)
Habits are acquisitions that require the use of reason and active
1 (Abow itz 2011) Education Education Acquisition Acquisition
preference.
Habits are developed dispositions for established forms of action and
2 (Abow itz 2011) Education Education Disposition Tendency
thought.
3 (Abow itz 2011) Education Education Habit is a form of executive skill, an efficiency in doing. Skill Ability
According to this view , habits are behaviors that are (or w ere)
functional in terms of achieving particular goals and strong
4 (Abraham & Sheeran 2003) Psychology Psychology Behavior Behavior
associations have developed betw een particular environmental cues
and particular action schemas.
Habits are socially shaped, unconscious modes of response to the
5 (Allen 2010) Literature Humanities Modes Ways
environment.
Habits do not just denote the general predictability of a particular
6 (Allen 2010) Literature Humanities individual's behavior, so that one can judge an act as "in character" or Predictability Predictability
"out of character."
Habits are determinant characteristics of most humans, and often
Business And
7 (Alshuw aikhat & Nkw enti 2003)
Economics
Business and economics reinforced by fear of the unknow n in the face of apparent or implied Characteristics Characteristics
change.
Negative habits are defined as thoughts, emotions, or behavioral
8 (Anshel & Kang 2007) Health Health and medical tendencies that the individuals perceive as undesirable or not in their Tendency Tendency
best interests.
9 (Anshel & Kang 2007) Health Health and medical habits are thoughts, emotions, or tasks experienced regularly Emotions Emotions
Habits are largely unconscious behaviors, so it is very difficult for us
10 (Arbuthnott 2009) Education Education Behavior Behavior
to even notice them.
Here, "habit" is used to embrace cognate terms (habitus, customary
behavior, habitual and routinized action). Habit denotes w hat William
11 (Archer 2010) Social Sciences Social Sciences Sequence Sequence
James termed "sequences of behavior that have become virtually
automatic"
The term 'habit' generally denominates a more or less self-actuating
12 (Archer 2010) Social Sciences Social Sciences disposition or tendency to engage in a previously adopted or acquired Tendency Tendency
form of action
The compromise concept of a reflexive habitus elides tw o concepts
that Bourdieu consistently distinguished: the semi-unconscious
13 (Archer 2010) Social Sciences Social Sciences Disposition Tendency
dispositions constituting habitus and reflexivity as self-aw areness of
them.
14 (Armstrong 1986) Semiotics Humanities Here habits are tendencies to act and products of experimentation Tendency Tendency
A habit is a kind of mental formula that predicts how one w ill act and
15 (Aydin 2009) Philosophy Humanities Formula Formula
w ish to act under certain conditions.
Habit is a w illingness to act, a proposition can be understood as a
16 (Aydin 2009) Philosophy Humanities Formulation Formula
formulation of a w illingness to act.
Habit is a repeated behavioral pattern diat automatically occurs
17 (Bansal 2011) Information Systems Science and Technology Pattern Pattern
w ithout one's conscious aw areness
Actually habits are not initiating forces in themselves; habits are
18 (Bayton 1957) Marketing Business and economics repeated response patterns accompanied by a minimum of cognitive Pattern Pattern
activity.
Habits are defined as assembled routines that link sensory cues to
19 (Berardelli et al. 2003) Neurology Health and medical Routines Routines
motor action.
20 (Berk & Galvan 2009) Social Sciences Social Sciences Modes of response. Modes Ways
21 (Berk & Galvan 2009) Social Sciences Social Sciences Acquired predisposition to w ays. Predisposition Tendency
22 (Berk & Galvan 2009) Social Sciences Social Sciences Habit is a repertoire. Repertoire Repertoire
Habit is defined by as a goal-directed response that persists after the
23 (Berridge 2007) Psychopharmacology Health and medical Response Response
goal itself
On this view habits are not patterns of action, but predispositions to
24 (Biesta 2007) Education Education Predisposition Tendency
act.
Habit is defined as behavior that has become automatic and is beyond
25 (Bonne et al. 2007) Psychology Psychology Behavior Behavior
an individual's aw areness.
Behavioral and Brain Habits are assembled routines that link sensory cues (either external
26 (Boyer & Linard 2006) Health and medical Routines Routines
Sciences or internal) w ith motor actions.
Dispositions or mental habits are formed after repetitious and
27 (Brand 2009) Social Sciences Social Sciences resembling actions but w e do not become curious by repeated acts of Disposition Tendency
curiosity.
habits are practices involving skill of sensory and motor organs,
28 (Brinkmann 2007) Philosophical Psychology Psychology Practices Routines
cunning or craft, and objective materials

The w ord habit is etymologically derived from the Latin verb habere ,
29 (Brockelman 2002) Health Health and medical
w hich in one derivation means "to hold" or "grasp," or "possess"
To Grasp Acquisition

The w ord habit is etymologically derived from the Latin verb habere ,
30 (Brockelman 2002) Health Health and medical
w hich in one derivation means "to hold" or "grasp," or "possess"
To Possess Acquisition

(Source:Author)

214

 Standardized
(Plurals and
No Reference Specific Discipline Grouped Discipline Definition Grouped as:
Regionalism s
elim inated)
From a phenomenological and existential point of view , habits are
31 (Brockelman 2002) Health Health and medical w ays in w hich humans shape their everyday behavior and attitudes Ways Ways
tow ard life into predictable patterns.

The w ord habit is etymologically derived from the Latin verb habere ,
32 (Brockelman 2002) Health Health and medical
w hich in one derivation means "to hold" or "grasp," or "possess".
To Hold Acquisition

Whereas habits are behaviors that have usually been often repeated
33 (Brder & Schiffer 2006) Psychology Psychology until they are so ingrained that they are automatically triggered by the Behavior Behavior
appropriate stimulus, routines do not require repetition to be learned.

34 (Bula 2004) Information Systems Science and Technology A habit is a process. Process Process
Habit is the tendency to respond in an automatic fashion to external or
35 (Canin, Dolcini & Adler 1999) Psychology Psychology Tendency Tendency
internal stimuli.
A persons preferences and habits are learned behavior and/or
36 (Carter & Fourney 2004) Information Systems Science and Technology Behavior Behavior
reflective of the personality and the mental model of the user.
Habits are patterned predispositions that enable people to respond to
37 (Charmaz 2002) Health Health and medical Predisposition Tendency
their situations w ith economy of thought and action.
Habits are typically considered to be patterned behavior or tasks,
38 (Crepeau 2007) Health Health and medical Behavior Behavior
habitual w ays of thinking are also important to understand.
Habits are typically considered to be patterned behavior or tasks,
39 (Crepeau 2007) Health Health and medical Ways Ways
habitual w ays of thinking are also important to understand.
40 (Crissman 1942) Psychology Psychology Responses w hich express a w ay of behaving. Predisposition Tendency
41 (Cutchin 2000) Health Health and medical Habit is an acquired predisposition to w ays or modes of response. Predisposition Tendency
Habits are modifiable structures of action that serve as the basis for
42 (Cutchin 2000) Health Health and medical Structures Structures
an intelligent resolution of contextual problems.
43 (Cutchin 2007) Health Health and medical Habits are the "ground-pattern" for all action and feeling. Pattern Pattern
44 (Cutchin 2007) Health Health and medical Habit is as a functional tool. Tool Tool
Habits are dispositions and tendencies that are specific to a set of
45 (Davis 2001) Psychology Psychology Disposition Tendency
stimuli and that guide behavior.
Habits are tendencies that are specific to a set of stimuli and that
46 (Davis 2001) Psychology Psychology Tendency Tendency
guide behavior.
47 (de Nooijer, Onnink & van Assema 2010) Health Health and medical Habits are automatic responses to specific cues. Response Response
Habits are patterns or w ays of dealing w ith external forces and
48 (Dobbs-Allsopp 2005) Religions And Theology Humanities Pattern Pattern
organizing internal energies in our interplay w ith the environment.
Habits are patterns or w ays of dealing w ith external forces and
49 (Dobbs-Allsopp 2005) Religions And Theology Humanities Ways Ways
organizing internal energies in our interplay w ith the environment.
Library And Information Habits are behaviors w e exhibit reliably on appropriate occasions and
50 (Donham 2010) Science and Technology Behavior Behavior
Sciences they are smoothly triggered w ithout painstaking attention.
51 (Dunn 2000) Health Health and medical Habits are patterns of human behavior. Pattern Pattern
52 (Fischbein 2011) Psychology Psychology For Peirce habit is a rule of action. Rule Formula
On this view , habit is a quasi-natural disposition insofar as it is
53 (Forman 2010) Philosophy Humanities Disposition Tendency
durable, brings pleasure, etc.
Habit is the most essential feature of the existence of all mental life in
54 (Forman 2010) Philosophy Humanities Feature Characteristics
the individual subject.
A habit is defined as a psychological construct implying goal directed
55 (Fujii & Kitamura 2003) Transportation Business and economics Construct Construct
automaticity in implementing a behavior.
56 (Garrison 2002) Health Health and medical Habits are functions. Functions Functions
57 (Garrison 2002) Health Health and medical Habits are means to ends. Means Means
Habits predispositions to respond the same w ay given the same
58 (Garrison 2002) Health Health and medical Predisposition Tendency
stimulus.
Habits are our tools and if w e haven't our kit of tools w ith us w e are
59 (Garrison 2002) Health Health and medical Tool Tool
certainly helpless.
Habits are embodied dispositions to act in a manner that manifests our
60 (Garrison 2003) Education Education Disposition Tendency
emotions.
Habit is an ability, an art, formed through past experience. But
w hether an ability is limited to repetition of past acts adapted to past
61 (Granger 1998) Education Education Ability Ability
conditions or is available for new emergencies depends w holly upon
w hat kind of habit exists.
62 (Granger 1998) Education Education Habits are immediate responses to situations. Response Response
Habits are tools that are a necessary part of everyday life, and can
63 (Granger 1998) Education Education Tool Tool
be both positive and negative, helpful and harmful.
64 (Greenberg 2009) Literature Humanities Habit is a dynamic force rather than an archive. Force Strength
Stein's habit is a gradualist and incremental but unpredictable and
65 (Greenberg 2009) Literature Humanities lively pattern of repetition w ith difference, a pattern that is "not Pattern Pattern
inevitable or uniform"
Habit is a predisposition to become involved in previously adopted or
66 (Guerreiro, Pereira & Frezatti 2006) Business Business and economics Predisposition Tendency
acquired forms of action.
Habit is a tendency to become involved in previously adopted or
67 (Guerreiro, Pereira & Frezatti 2006) Business Business and economics Tendency Tendency
acquired forms of action.
68 (Hedoin 2009) Economics Business and economics Habit is a behavioral predisposition. Predisposition Tendency
We treat habits and routines as dispositions, rather than expressed
69 (Hodgson, Geoffrey & Thorbjrn 2006) Evolutionary Economics Business and economics Disposition Tendency
behavior as such.
70 (Hodgson, Geoffrey & Thorbjrn 2006) Evolutionary Economics Business and economics Habits and routines are replicators. Replicators Replicators

(Source:Author)

215


Standardized
(Plurals and
No Reference Specific Discipline Grouped Discipline Definition Grouped as:
Regionalism s
elim inated)
Behavioral and Brain
71 (Hodgson, Geoffrey 2007) Health and medical Both instincts and habits are rule-like dispositions. Disposition Tendency
Sciences
Habit is a disposition to engage in previously adopted or acquired
72 (Hodgson, Geoffrey 2009) Economics Business and economics Disposition Tendency
behavior.
73 (Hodgson, Geoffrey 2009) Economics Business and economics Habits are the preconditions for all reason and deliberation. Precondition Precondition
74 (Hodgson 2010) Economics Business and economics Habits are submerged repertoires. Repertoire Repertoire
75 (Kemp 1998) History of Psychology Psychology Habits are enduring dispositions. Disposition Tendency
Habit is a special tendency by virtue of w hich w hat has been done
76 (Ku 2007) Semiotics Humanities Tendency Tendency
w ill be done oftener than w hat has not been done.
Psychiatry And A habit is an automatic action in a given situation, w ithout direct
77 (Law rence, Evans & Lees 2003) Health and medical Action Behavior
Neurology reference to the goal of that action.
For habits are tendencies, not law s, w hich is w hy chance may
78 (Lefebvre 2007) Philosophy Humanities Tendency Tendency
manifest itself in agapastic evolution.
Habit is an example of irrational behavior because an individual
79 (Liao et al. 2011) Information Systems Science and Technology continues to do w hat they are habitually used to doing w ithout Behavior Behavior
applying rational analysis to the behavior.
Habit is the inclination for behavior and it represents current
80 (Liao et al. 2011) Information Systems Science and Technology Inclination Tendency
behavioral preference.
81 (Liberman & Frster 2008) Psychology Psychology Habit is the strength of association betw een action and outcome. Strength Strength
Habits are ripples, bits of behavior that form routines w hen strung
82 (Luebben & Royeen 2007) Health Health and medical Ripples Ripples
together into a sequence.
83 (MacMullan 2005) Philosophy Humanities Habit is an acquired predisposition to w ays or modes of response. Predisposition Tendency
Learned sequence of acts that have become automatic responses to
84 (Mair & Bergin-seers 2010) Tourism and Hospitality Business and economics specific cues and are functional in obtaining certain goals or end Sequence Sequence
states.
Habits are pow erful forces. Much like gravity, habits exert a
85 (McGuinness & McElroy 2010) Social Psychology Psychology Force Strength
tremendous force and are difficult to change.
Habits are dispositions and tendencies that are specific to a set of
86 (Meyer & Sugiyama 2007) Know ledge Management Business and economics Disposition Tendency
stimuli and guide behavior.
Habits are tendencies that are specific to a set of stimuli and guide
87 (Meyer & Sugiyama 2007) Know ledge Management Business and economics Tendency Tendency
behavior.
An acquired predisposition to w ays/modes of response, a sensibility
88 (Miller 2010) Business Business and economics to certain classes of stimuli, and a reference to one's standing Predisposition Tendency
predilections and aversions.
Habits are automatic behavioral routines developed w ithin a
consistent environment, such that changing the environment may
89 (Moore et al. 2006) Health Health and medical Routines Routines
encourage conscious decision-making and therefore a change in
habit.
90 (Moses 2009) Literature Humanities Habits are proclivities. Proclivity Tendency
91 (Moses 2009) Literature Humanities Habits are tendencies. Tendency Tendency
Habits are also tastes - habitual modes of preference and esteem, an
92 (Nakamura 2009) Education Education Modes Ways
effective sense of excellence.
Habits are also tastes - habitual modes of preference and esteem, an
93 (Nakamura 2009) Education Education Tastes Preference
effective sense of excellence.
94 (Nslindh-ylispangar et al. 2008) Health Health and medical Habit is a learned behavior. Behavior Behavior
Habit is a tool for identifying generic patterns in the situation-specific
95 (Norros & Salo 2009) Cognition & Technology Science and Technology Tool Tool
behavior of the system.
Habits are actions that have become embodied in the nervous system
96 (Otten 1999) Literature Humanities Action Behavior
until w e don't think about them anymore.
Habit is by definition an unconscious acquisition and tends to become
97 (Pasticci 2009) Music Humanities Acquisition Acquisition
mechanical.
Habits are rhythmic and consisting of patterns that w ere alw ays
98 (Peng et al. 2011) Criminology Science and Technology Pattern Pattern
repeated.
In this view , nature is the remote source, and habits are the proximate
99 (Petillo 2010) Theological Studies Humanities Source Source
source, of acts that attain a kind of natural beatitude.
100 (Poole 2000) Health Health and medical Habits are patterns that organize daily life Pattern Pattern
Consumer purchasing habits are patterns of consumer choice
101 (Pow ers & Loyka 2007) Marketing Business and economics outcomes for products bought and may be unique to specific regions Pattern Pattern
or markets.
Habit is seen as a pattern of overt behavior extended in time rather
102 (Rachlin 2002) Psychology Psychology Pattern Pattern
than, as sometimes seen in psychology, as an internal state.
103 (Ralston 2011) Education Education Habit is a manner of action, not a particular act or deed. Manner Ways
104 (Ralston 2011) Education Education Habit is a mode of conduct, not the conduct itself. Modes Ways
105 (Ralston 2011) Education Education A w ay of action, not a particular act or deed. Ways Ways
106 (Raposa 2006) Theology & Philosophy Humanities Habits are skills. Skill Ability
The essence of a habit is an acquired predisposition to w ays of
107 (Reynolds 1981) Economics Business and economics Predisposition Tendency
response.
Habits are conceived as behavioral responses brought on by
108 (Rhodes & de Bruijn 2010) Health Health and medical Response Response
environmental cues.
109 (Ronald Lee 1998) Education Education Predisposition to w ays or modes of response, not particular acts Predisposition Tendency
110 (Satyaprasad 2009) Health Health and medical Habits are learned patterns of muscular contraction. Pattern Pattern

(Source:Author)

216

Standardized
(Plurals and
No Reference Specific Discipline Grouped Discipline Definition Grouped as:
Regionalism s
elim inated)
111 (Schfer, Jaeger-erben & Bamberg 2012) Business Business and economics The terms routines and habits are used as synonyms in this article. Routines Routines

Repetition or habit is an individual process of automatic behavior

112 (Schmuck & Vlek 2003) Psychology Psychology w hich has been w ell established over time as the result of recurrent Repetition Repetition
positive reinforcements and the absence of major punishments.

113 (Schultz & Searleman 2002) Psychology Psychology A habit is a typical pattern of behavior. Pattern Pattern
Habits are operationally defined as behavior that continues although
114 (Schw abe, Dickinson & Wolf 2011) Psychiatry Health and medical the incentive value of the goal has been reduced in a devaluation Behavior Behavior
procedure.
Habits are dispositions and tendencies that are specific to a set of
115 (Seger 1994) Psychology Psychology Disposition Tendency
stimuli and that guide behavior.
Habits are tendencies that are specific to a set of stimuli and that
116 (Seger 1994) Psychology Psychology Tendency Tendency
guide behavior.
Social Services And
117 (Seton 2001) Social Sciences Habits are templates hard to relinquish. Templates Replicators
Welfare
118 (Shields 2008) Public Administration Business and economics Habits are solutions in practice to past problematic situations. Solutions Solutions
Habit is that of a self-analyzing and self-correcting disposition to act in
119 (Singer 1981) Semiotics Humanities Disposition Tendency
a certain w ay under given circumstances and motivation.
120 (Sonnentag & Jelden 2009) Health Health and medical Habit is a behavioral tendency to repeat responses. Tendency Tendency
Habits are not merely practices that issue from resolved thought;
121 (Stengel 2010) Philosophy Humanities rather, habits are the resolution of idea, disposition or act, and affect. Disposition Tendency
Habits hold affect in relation to act and idea.
Habits are not merely practices that issue from resolved thought;
122 (Stengel 2010) Philosophy Humanities rather, habits are the resolution of idea, disposition or act, and affect. Resolution Solutions
Habits hold affect in relation to act and idea.
A socially significant form of human behavior and none discuss it
123 (Sw artz 2002) Health Health and medical Behavior Behavior
extensively.
a system of durable, transposable dispositions, structured structures
124 (Sw artz 2002) Health Health and medical System System
predisposed to function as structuring structures.
Among other things these tendencies/habits and behaviors are
125 (Sw eet, Roome & Sw eet 2003) Business Business and economics Tendency Tendency
sensitive to self-consciousness of our normative surroundings.
Voluntary and Non-profit Habits are automatic responses of individuals to specific cues. They
126 (Theuvsen 2004) Business and economics Response Response
Organizations reflect a behavioral tendency to repeat responses.
Learned sequences of acts that have become automatic responses to
127 (Thgersen & Mller 2008) Transportation Business and economics Sequence Sequence
specific cues, and are functional in obtaining certain goals.
Habits are defined as slow ly developing associations betw een
128 (Tobias 2009) Psychology Psychology Association Association
situational cues and repeatedly performed behavior options.
Habit is a recurrent, often unconscious, behavioral pattern.
129 (van Berkel et al. 2011) Health Health and medical Pattern Pattern

Habits are defined as behavior patterns that become regular or

130 (Vandenberg 2010) Psychology Psychology Pattern Pattern
spontaneous due to regular repetition.
These habits are patterns of action acquired by training that uses
131 (Velamuri & Dew 2010) Ethics Humanities Pattern Pattern
pleasure and pain as incentives
Applying this insight to habits, habit is behavior that has a history of
repetition, is characterized by a lack of aw areness and conscious
132 (Verplanken et al. 2007) Social Psychology Psychology Behavior Behavior
intent, is mentally efficient, and is sometimes difficult to control. In
addition, habits may be part of a person's
Habit is both a system of dispositions to action and a repository of the
133 (Wozniak 2009) Psychology Psychology Repository Repository
effects of the organisms past experience.
Habit is both a system of dispositions to action and a repository of the
134 (Wozniak 2009) Psychology Psychology System System
effects of the organisms past experience.
Habits are situation-behavior sequences that are or have become
135 (Yoon 2011) Business Business and economics Sequence Sequence
automatic and that occur w ithout self-instruction.

(Source:Author)

217

 APPENDIX2SAMPLEOFDEFINITIONSBYDISCIPLINE

Asampleofdefinitionsofhabit(s)wereextractedfrom5532recordsofpeerreviewedarticlescontainingthekeyphrases:habit
isorhabitsareinthe43databasesofProQuestCentral(Feb2012).Fromthe5532only144articlescontainedactualdefinitions
ofhabit,anditsdistributionbydisciplineisshownhere.
NUMBEROF NUMBEROF
DISCIPLINE DEFINITIONS % SPECIFICDISCIPLINE DEFINITIONS REFERENCE
Business 6
BusinessandEconomics 1
Economics 5
(Alshuwaikhat & Nkwenti 2003; Bayton 1957; Fujii & Kitamura
EvolutionaryEconomics 2 2003; Guerreiro & Frezatti 2006; Hedoin 2009; Hodgson 2010;
KnowledgeManagement 2 Hodgson & Thorbjrn 2006; Hodgson 2009; Mair & Berginseers
Businessand
23 Marketing 2 2010;Meyer&Sugiyama2007;Miller2010;Powers&Loyka2007;
economics
Reynolds 1981; Schfer, Jaegererben & Bamberg 2012; Shields
PublicAdministration 1 2008; Sweet, Roome & Sweet 2003; Theuvsen 2004; Thgersen &
TourismandHospitality 1 Mller2008;Yoon2011)
Transportation 2
VoluntaryandNonprofit
1
Organizations
(Abowitz 2011; Arbuthnott 2009; Biesta 2007; Garrison 2003;
Education 15 Education 15
Granger1998;Nakamura2009;Ralston2011;Ronald1998)

BehavioralandBrainSciences 2 (Anshel&Kang2007;Berardellietal.2003;Berridge2007;Boyer&
Linard 2006; Brockelman 2002; Charmaz 2002; Crepeau 2007;
Health 29 Cutchin2000,2007;deNooijer,Onnink&vanAssema2010;Dunn
Healthand Neurology 1 2000;Garrison2002;Hodgson2007;Lawrence,Evans&Lees2003;
35
medical Luebben&Royeen2007;Mooreetal.2006;Nslindhylispangaret
Psychiatry 1 al.2008;Poole2000;Rhodes&deBruijn2010;Satyaprasad2009;
PsychiatryAndNeurology 1 Schwabe,Dickinson&Wolf2011;Sonnentag&Jelden2009;Swartz
2002;vanBerkeletal.2011)
Psychopharmacology 1
Ethics 1
Literature 7
(Allen 2010; Armstrong 1986; Aydin 2009; DobbsAllsopp 2005;
Music 1 Forman2010;Greenberg2009;Ku2007;Lefebvre2007;MacMullan
Humanities 24
Philosophy 8 2005;Moses2009;Otten1999;Pasticci2009;Petillo2010;Raposa
2006;Singer1981;Stengel2010;Velamuri&Dew2010)
ReligionsAndTheology 4
Semiotics 3
HistoryofPsychology 1 (Abraham & Sheeran 2003; Bonne et al. 2007; Brinkmann 2007;
PhilosophicalPsychology 1 Brder&Schiffer2006;Canin,Dolcini&Adler1999;Crissman1942;
Davis2001;Fischbein2011;Kemp1998;Liberman&Frster2008;
Psychology 22 Psychology 18 McGuinness&McElroy2010;Rachlin2002;Schmuck&Vlek2003;
Schultz & Searleman 2002; Seger 1994; Tobias 2009; Vandenberg
SocialPsychology 2
2010;Verplankenetal.2007;Wozniak2009)
CognitionandTechnology 1
Scienceand Criminology 1 (Bansal 2011; Bula 2004; Carter & Fourney 2004; Donham 2010;
8
Technology InformationSystems 5 Liaoetal.2011;Norros&Salo2009;Pengetal.2011)
LibraryAndInformationSciences 1
Social SocialSciences 7
8 (Archer2010;Berk&Galvan2009;Brand2009;Seton2001)
Sciences SocialServicesAndWelfare 1
Total 135 100% 135

(Source:Author)

218

 APPENDIX3COREHABITDEFINITIONSGROUPED

Cambridge Dictionary online (2012) Oxford Dictionary online (2012)

Group KEY WORD
http://dictionary.cambridge.org http://oxforddictionaries.com
the process of doing something, especially when the fact or process of doing something, typically to
Action dealing with a problem or difficulty achieve an aim
(to act) [How someone acts] in a particular way, or
the way in which one acts or conducts oneself,
1
Behavior to be good by acting in a way which has society's
especially towards others
approval
Movements a change of position an act of moving
a piece of work to be done, especially one done
Tasks a piece of work to be done or undertaken
regularly, unwillingly or with difficulty
a group of people who work together in a single
Association a mental connection between things
organization for a particular purpose
2
a relationship between two things or situations,
Link a connection between two people, things or ideas
especially where one affects the other
to arrange something or a collection of things so a performance, show, or event staged for public
Display that they can be seen by the public entertainment
a stock of plays, dances, or items that a company
3
all the music or plays, etc. that you can do or or a performer knows or is prepared to perform, a
Repertoire perform or that you know stock of skills or types of behaviour that a person
habitually uses
the particular type of character which a person
the way in which something is placed or arranged,
Disposition naturally has or a natural tendency to do
especially in relation to other things
something, or to have or develop something
a preference or tendency, or a feeling that makes a a persons natural tendency or urge to act or feel in
Inclination person want to do something a particular way; a disposition
a liability or tendency to suffer from a particular
the state of being likely to behave in a particular
Predisposition way or to suffer from a particular disease condition, hold a particular attitude, or act in a
particular way
4 a tendency to choose or do something regularly; an
a tendency to do or like something, especially
Proclivity something against moral laws
inclination or predisposition towards a particular
thing
a tendency towards a particular way of behaving, an inclination or natural tendency to behave in a
Propensity especially a bad one particular way
If someone has a tendency to do or like something,
they will probably do it or like it, If there is a an inclination towards a particular characteristic or
Tendency tendency for something to happen, it is likely to type of behavior
happen or it often happens
the physical or mental power or skill needed to do
Ability something
an activity through which people express particular a skill at doing a specified thing, typically one
Arts ideas acquired through practice
5 the total amount that can be contained or produced,
Capacity or (especially of a person or organization) the ability the ability or power to do or understand something
to do a particular thing
an ability to do an activity or job well, especially
Skill the ability to do something well; expertise
because you have practised it
(Source:Author)

219

 Cambridge Dictionary online (2012) Oxford Dictionary online (2012)
Group KEY WORD
http://dictionary.cambridge.org http://oxforddictionaries.com
Acquisition the process of getting something
to quickly take something in your hand(s) and hold it
ToGrasp firmly
seize and hold firmly;
6 grasp, carry, or support with ones arms or hands;
ToHold to cause someone to act on a promise or agreement
keep or detain
to have or own something, or to have a particular
ToPossess quality
have as belonging to one; own

a typical or noticeable quality of someone or a feature or quality belonging typically to a person,

Characteristics something place, or thing and serving to identify them
7
Feature a typical quality or an important part of something a distinctive attribute or aspect of something
a particular characteristic that can produce a a distinguishing quality or characteristic, typically
Traits particular type of behaviour one belonging to a person
an act or the possibility of choosing (to decide what an act of choosing between two or more
Choice you want from two or more things or possibilities) possibilities

when you like something or someone more than a greater liking for one alternative over another or
8 Preference another person or thing others
a person's ability to judge and recognise what is
a persons tendency to like or be interested in
Tastes good or suitable, especially relating to such matters
something
as art, style, beauty and behaviour
a standard or accepted way of doing or making
something, the things needed for it, or a a mathematical relationship or rule expressed in
Formula mathematical rule expressed in a set of numbers symbols; a rule or style followed mechanically
and letters
9 to develop all the details of a plan for doing a material or mixture prepared according to a
Formulation something formula
an accepted principle or instruction that states the one of a set of explicit or understood regulations or
Rule way things are or should be done, and tells you principles governing conduct or procedure within a
what you are allowed or are not allowed to do particular area of activity;
when you solve or end a problem or difficulty; a a firm decision to do or not to do something; the
Resolution promise to yourself to do or to not do something action of solving a problem or contentious matter
10
a means of solving a problem or dealing with a
Solutions the answer to a problem
difficult situation
the way in which something is done; in the style of
Manner a way in which a thing is done or happens
something
a way or manner in which something occurs or is
Modes a way of operating, living or behaving
11 experienced, expressed, or done
a route, direction or path, used to talk about the
a method, style, or manner of doing something; an
Ways direction in which something is facing, used in the
optional or alternative form of action
names of some roads
in scientific use, (a measure of) the influence which strength or energy as an attribute of physical action
Force changes movement or movement
12 the quality or state of being physically strong; the
the ability to do things that need a lot of physical or
Strength mental effort
capacity of an object or substance to withstand
great force or pressure;
(Source:Author)

220

 Cambridge Dictionary online (2012) Oxford Dictionary online (2012)
Group KEY WORD
http://dictionary.cambridge.org http://oxforddictionaries.com
[replicate] to make or do something again in
Replicators a thing which replicates or copies something
exactly the same way
13 a shaped piece of rigid material used as a pattern
something that is used as a pattern for producing
Templates other similar things
for processes such as cutting out, shaping, or
drilling
the actual application or use of an idea, belief, or
something that is usually or regularly done, often as method, as opposed to theories relating to it; the
Practices a habit, tradition or custom customary, habitual, or expected procedure or way
14 of doing of something
a usual or fixed way of doing things; a regular
Routines series of movements, jokes or similar things used a sequence of actions regularly followed
in a performance
working by itself with little or no direct human
An automatic machine or device is able to operate
15 Automaticity independently of human control
control, done or occurring spontaneously, without
conscious thought or attention
an idea or theory containing various conceptual
to build something or put together different parts to
16 Construct form something whole
elements, typically one considered to be subjective
and not based on empirical evidence
a strong feeling such as love or anger, or strong a strong feeling deriving from ones circumstances,
17 Emotions feelings in general mood, or relationships with others
the natural purpose (of something) or the duty (of a
the natural purpose (of something) or the duty (of a
18 Functions person)
person); a thing dependent on another factor or
factors
an action or system by which a result is achieved;
19 Means a method or way of doing something
a method

20 Mindset a person's way of thinking and their opinions the established set of attitudes held by someone

a regular and intelligible form or sequence

discernible in the way in which something happens
a particular way in which something is done, or is done; Origin:
21 Pattern organized or happens Middle English patron 'something serving as a
model', from Old French (see patron). The change
in sense is from the idea of a patron giving a
something that exists and can be seen, felt, tasted,
a fact or situation that is observed to exist or
22 Phenomenon etc., especially something which is unusual or
happen, especially one whose cause or explanation
interesting
something which must happen or be true before it is a condition that must be fulfilled before other things
23 Precondition possible for something else to happen can happen or be done
able to be predicted [say or estimate that (a
the state of knowing what something is like, when
24 Predictability something will happen
specified thing) will happen in the future or will be a
consequence of something]
a series of actions that you take in order to achieve a series of actions or steps taken in order to
25 Process a result achieve a particular end
(Source:Author)

221

 Cambridge Dictionary online (2012) Oxford Dictionary online (2012)
Group KEY WORD
http://dictionary.cambridge.org http://oxforddictionaries.com
something that happens in the same way as
26 Repetition the recurrence of an action or event
something which happened before
a place where things are stored and can be found; a
a place where or receptacle in which things are or
27 Repository person who has, or a book that contains, a lot of
may be stored
information or detailed knowledge
28 Response an answer or reaction a reaction to something
a small wave or series of waves on the surface of
29 Ripples a small wave on the surface of water water, especially as caused by a slight breeze or
an object dropping into it
handwriting as distinct from print; written
30 Script the words of a film, play, broadcast or speech characters, an automated series of instructions
carried out in a specific order
a series of related things or events, or the order in a particular order in which related things follow each
31 Sequence which they follow each other other
the place something comes from or starts at, or the
cause of something; someone or something that a place, person, or thing from which something
32 Source supplies information; at the place where something originates or can be obtained;
comes from
the way in which the parts of a system or object are
the arrangement of and relations between the parts
33 Structures arranged or organized, or a system arranged in this
or elements of something complex
way
a set of things working together as parts of a
mechanism or an interconnecting network; a
a set of connected things or devices which operate
34 System together
complex whole; a set of principles or procedures
according to which something is done; an
organized scheme or method
a piece of equipment which you use with your
a device or implement, especially one held in the
35 Tool hands to make or repair something; something that
hand, used to carry out a particular function
helps you to do a particular activity
(Source:Author)

222

 APPENDIX4TYPEOFHABITMEASURED

Singlepredeterminedhabit(SP)
Multiplepredeterminedhabits(MP)
Singlesemipredetermined(SS)
Multiplesemipredetermined(MS)
Singlenonpredeterminedhabit(SN)
Multiplenonpredeterminedhabits(MN)

MN(6)
MP(2)

MS(4)

SN(5)
SP(1)

SS(3)
No. Reference Behaviorofthestudy Notes

(EscobarRodrguez&CarvajalTrujillo
1 Onlineairlineticketpurchasing 1
2013)
2 (Han&Farn2013) PervasiveBusinessIntelligenceSystems 1

3 (Huang,Wu&Chou2013) Dataminingtools 1

4 (Kangetal.2013) Socialnetworksites 1
5 (Klckner2013) Environmentallyrelevantbehavior 1
6 (Nikou&Bouwman2013) MobileSocialNetwork 1

7 (Raman&Don2013) asLearningManagementSystem 1

8 (Tseng,Chang&Woo2013)
9 (Venkatesh,Thong&Xu2012)
Alltheitemsrefertocaruse(singlehabit).Theitemsare
Likertscales.ThestudyclaimshaveusedRFM,butitisa
Drivingbehavior 1
significantvariation.Itusesdiversesituations,all
evaluatingtheuseofcarsonly.
mobileInternettechnology 1

10 (Barnes2011) Usecontinuanceinvirtualworlds 1

Pushstrategiestoreducetheusagedemand
11 (Chen&Lai2011) ofmotorizedvehicles,andpullstrategyto
attractmorepublictransportusers
7itemsrepresentimaginarysituationsthatrequire
traveling.Therespondenthas3optionsineachcase:
motorcycle,carorpublictransport.Eachhabitis
1
assessedbythenumberoftimesitwasselected.
(Althoughtheinstrumentsuggestvariousanswersare
possiblesimultaneously,onlyoneis)

12 (Loibl,Kraybill&DeMay2011) Savingshabitsinregularsaving 1

Bingedrinkingamongundergraduate
13 (Norman2011) 1
students
14 (Pahnila,Siponen&Zheng2011) UseofChineseebay 1

15 (DeBruijn&Rhodes2010) Exercisehabit 1

16 (Guetal.2010) Usersdriverstoclickads 1
Studentsperceptionsanduseofauniversity
17 (Lankton,Wilson&Mao2010) 1
internetapplication(UIA)
18 (deBruijnetal.2009) Adultbicycleuseasameanstransportation 1

Adolescentsoftdrinkconsumption,
19 (DeBruijn&VanDenPutte2009) 1
televisionviewingandhabitstrength

20 (Gardner2009) Travelmode 1

(Source:Author)

223

 MN(6)
MP(2)

MS(4)

SN(5)
SP(1)

SS(3)
No. Reference Behaviorofthestudy Notes

21 (Limayem,Moez&Cheung2008) BlackboardLearningSystem 1
22 (Wu&Kuo2008) Googlesearchengine 1
23 (DeBruijnetal.2007) Fruitconsumption 1
24 (Limayem,Hirt&Cheung2007) VoluntarycontinuedWWWusage 1
25 (Liao,Palvia&Lin2006) Websiteuseandecommerceadoption 1
26 (Thgersen2006) TravelModeChoice 1
27 (vanEmpelen&Kok2006) Useofcondom 1

Eating,mentalhabitsandword
28 (Verplanken2006) 1
processing

29 (Honkanen,Olsen&Verplanken2005) Seafoodconsumptionbehaviour 1
30 (Kim&Malhotra2005) Website 1
31 (Wood,Tam&Witt2005) Exercise,readingnewspaper,watchingTV 1

32 (Gefen2003) B2Cwebsite 1

SpecificRFMmeasuredSP,abipolarmeasureMP,a
33 (Klckner,Matthies&Hunecke2003) TravelModeChoice 1 1 1 1 slightlyadaptationtothe"original"RFMSS,multiple
RFMMS.
34 (Limayem,&Hirt2003) OReillysWebBoard3.5(Educationtool) 1
35 (Limayem,Cheung&Chan2003) BlackboardLearningSystem 1
36 (Limayem,Hirt&Cheung2003) WWW 1
Study1,2&3(SRHI=SP),Study4tookalistdailyand
37 (Verplanken,Bas&Orbell2003) Travelmodechoice 1 1 weeklyhabits,thehabitwiththehighestfrequencywas
selectedSN
Respondentsgetquestionsabouttenimaginary
38 (Bamberg&Schmidt2003) CarUse 1 situationsthatrequiretraveling.Theyhadtoindicateas
quicklyaspossiblethetravelmodetheywouldchoose.

39 (Orbelletal.2001) Ecstasyuse 1

40 (Saba,Vassallo&Turrini2000) Consumptionoffatcontainingfoods 1

41 (Trafimow2000) Useofcondom 1
Listwith67foods,37unhealthy.Participantschecked
productstheyconsumedinthelastweek.Thenumberof
42 (Verplanken,Bas&Faes1999) Unhealthyfoodhabits 1
unhealthyfoodswastakenasmeasureofunhelaty
habits.

43 (Ouellette&Wood1998) HabitandIntentioninEverydayLife 1

44 (Saba&DiNatale1998a) Oliveoil 1

45 (Saba&DiNatale1998b) Meatconsumption 1

46 (Sabaetal.1998) Milkconsumption 1

(Source:Author)

224

MN(6)
MP(2)

MS(4)

SN(5)
SP(1)

SS(3)
No. Reference Behaviorofthestudy Notes

47 (Verplanken,Basetal.1998)
48 (Bergeronetal.1995)
(Verplanken,Aarts,vanKnippenberg,&
49
vanKnippenberg1994)
15imaginarytrips,wereaskedtoindicateasquicklyas
TravelModeChoice 1
possiblewhattravelmodetheywouldchoose.
Executiveinformationsystems 1
10imaginarysituationcallingforachoiceoftravelmode
werepresented.Sixmodesweregivenaspossible
CarUse 1 choices(bicycle,bus,cab,car,trainandwalking).Car
choicehabitwascalculatedfromthenumberoftimesit
wasselected.

50 (Towler&Shepherd1992) Consumptionofahighfatfood(chips) 1

51 (Ajzen1991) Diversebehaviors(Aggregateprinciple)

52 (Montano&Taplin1991) Mammographyparticipation 1
53 (Bagozzi&Warshaw1990) Losingweight 1
54 (Charng,Piliavin&Callero1988) Blooddonation 1
55 (Mittal1988) SeatBeltUsage 1
56 (Wittenbraker,Gibbs&Kahle1983) UseofSeatBelt 1
57 (Bagozzi1981) Blooddonation 1

Frequencyobservation:(a)affectiveclarificationand
acceptance(totrust,tolove);(b)praiseandreward(to
admire,torespect);(c)cognitiveandskillclarification(to
discusswork,toworkwith);(d)correctivefeedback(totry
totalkhim/herintoownpointofview,tohelp);(e)
requestsandcommands(giveorderstohim/her,to
58 (Landis,Triandis&Adamopoulos1978) Classroomteacherbehavior 1
disciplinehim/her);(f)criticismandrejection(tocriticize,
totellhim/heroff);(g)laugh(laughtogether,playgames
with);(h)negative
physicalcontact(hithim/her,tothreatenhim/her);(f)
positivephysicalcontact(befriendswith,treathim/her
asabrother/sister).

(Source:Author)

225

 APPENDIX5EMPIRICALEVIDENCEOFTHERELATIONSHIPSOFHABIT

Food&Drinks

Regression
Transport
Exercise
Health

Other

Other
SEM
PLS
ICT
No. Reference Behaviorofthestudy Discipline Theory Measurement SampleSize AnalysisType BI AB M

(EscobarRodrguez (Limayem,Hirt& 1360adults(non

1 &CarvajalTrujillo Onlineairlineticketpurchasing IS 1 1 UTAUT2 Cheung2003) randomsample) 1 SEM Y Y
2013) (Adapted) (Diverseregions)
PervasiveBusinessIntelligence (Limayem,Hirt& 117students
2 (Han&Farn2013) IS 1 1 1 PLS Y N
Systems Cheung2003) (snowball)
(Huang,Wu&Chou (Limayem,Hirt&
3 Dataminingtools IS 1 1 TTF,ECM 285MBAalumni 1 PLS,SEM Y
2013) Cheung2003)
TRA,TPB,TAM,
UTAUT,EMC, (Limayem,Hirt&
4 (Kangetal.2013) Socialnetworksites IS 1 1 278students 1 PLS Y
Postadoption Cheung2003)
ITModel
TPB,NATor 56Research
5 (Klckner2013) Environmentallyrelevantbehavior Environment 1 1 1 MASEM Y
VBN reports
(Nikou&Bouwman (Nikou&Bouwman
6 MobileSocialNetwork IS 1 1 336users(China) 1 SEM N
2013) 2013)(Developed)
(Limayem,Hirt& 320students
7 (Raman&Don2013) asLearningManagementSystem IS 1 1 UTAUT2 1 PLS N N
Cheung2003) (Malaysia)
(Tseng,Chang& 544drivers
8 Drivingbehavior Transport 1 1 TPB RFM 1 Regression Y
Woo2013) (Taiwan)
(Venkatesh,Thong (Limayem,Hirt& 1.512users(Hong
9 mobileInternettechnology IS 1 1 UTAUT2 1 CMV,PLS,CFA Y Y
&Xu2012) Cheung2003) Kong)
(Limayem,Hirt&
10 (Barnes2011) Usecontinuanceinvirtualworlds IS 1 1 339users 1 PLS Y
Cheung2003)
Pushstrategiestoreducetheusage
231commuters
demandofmotorizedvehicles,and
11 (Chen&Lai2011) Transport 1 1 TPB RFM (Taipeiand 1 Regression Y
pullstrategytoattractmorepublic
Kaohsiung)
transportusers
269participants
(Loibl,Kraybill& SRHI(Verplanken,Bas Multiple
12 Savingshabitsinregularsaving Savings 1 1 TPB (treatmentgroup) 1 Y
DeMay2011) &Orbell2003) regression

137and109
Bingedrinkingamongundergraduate SRHI(Verplanken,Bas Hierarchical
13 (Norman2011) Health 1 1 TPB Undergraduate 1 Y
students &Orbell2003) regression
students
(Pahnila,Siponen& SRHI(Verplanken,Bas 180studentsin
14 UseofChineseebay IS 1 1 UTAUT 1 PLS,SEM Y
Zheng2011) &Orbell2003) USA
538 Descriptives,
(DeBruijn,G& SRHI(Verplanken,Bas undergraduate correlation,
15 Exercisehabit Exercise 1 1 TPB 1 Y
Rhodes2010) &Orbell2003) studentsinthe discriminant
Netherlands function
10,000randomly PLSPath
16 (Guetal.2010) Usersdriverstoclickads IS 1 1 Ownmodel Other 1 Y
sampledusers Modeling
371
(Lankton,Wilson& Studentsperceptionsanduseofa (Limayem,Hirt& ANOVA,PLS
17 IS 1 1 Ownmodel undergraduate 1 N
Mao2010) universityinternetapplication(UIA) Cheung2003) Graph
students
Hierarchical
(DeBruijn,GJ&Van Adultbicycleuseasameans SRHI(Verplanken,Bas 312Dutch regressionand
18 Transport 1 1 TPB 1 Y Y
DenPutte2009) transportation &Orbell2003) adolescents interaction
analysis
Hierarchical
(deBruijn,GJetal. Adolescentsoftdrinkconsumption, Foodordrinks SRHI(Verplanken,Bas
19 1 1 TPB 317Dutchadults 1 regression, Y Y
2009) televisionviewingandhabitstrength consumption &Orbell2003)
simpleslope
107staffand
SRHI(Verplanken,Bas Simpleslope,
20 (Gardner2009) Travelmode Transport 1 1 TPB studentcar 1 Y Y N
&Orbell2003) regression
commuters
(Limayem,Moez& (Limayem,Hirt& 505students
21 BlackboardLearningSystem IS 1 1 1 PLS Y Y
Cheung2008) Cheung2003) (USA)
SRHI(Verplanken,Bas 232convenience
22 (Wu&Kuo2008) Googlesearchengine IS 1 1 TAM 1 PLS Y Y
&Orbell2003) sample
CFA,multi
(DeBruijn,GJetal. Foodordrinks SRHI(Verplanken,Bas
23 Fruitconsumption 1 1 TPB 521Dutchadults 1 grouppath Y
2007) consumption &Orbell2003)
analyses
(Elementsfrom 553and227
(Limayem,M.,Hirt& (Limayem,Hirt&
24 VoluntarycontinuedWWWusage IS 1 1 TAMand university 1 PLS Y Y
Cheung2007) Cheung2003)
others) students
(Liao,Palvia&Lin Websiteuseandecommerce (Gefen2003) 446studentsand
25 IS 1 1 1 SEM,CFA Y
2006) adoption (Adapted) employees

(Source:Author)

226

 Food&Drinks

Regression
Transport
Exercise
Health

Other

Other
SEM
PLS
ICT
No. Reference Behaviorofthestudy Theory Measurement SampleSize AnalysisType BI AB M

924consumers
26 (Thgersen2006) TravelModeChoice Transport 1 1 TPB Pastbehavior 1 SEM Y
(Denmark)
(vanEmpelen&Kok (vanEmpelen&Kok 399students(The
27 Useofcondom Health 1 1 1 SEM Y Y
2006) 2006)(Developed) Netherlands)
Frequency,SRHI
Eating,mentalhabitsandword 128,194,76
28 (Verplanken2006) Various 1 1 TPB (Verplanken,Bas& 1 Regression Y
processing students
Orbell2003)
(Honkanen,Olsen& Foodordrinks SRHI(Verplanken,Bas 1579adults
29 Seafoodconsumptionbehaviour 1 1 TRA/TPB 1 SEM Y Y
Verplanken2005) consumption &Orbell2003) (Norway)
TAM,belief
(Kim&Malhotra updating,self 189students
30 Website IS 1 1 Pastbehavior 1 SEM Y
2005) perception, (USA)
habit
(Wood,Tam&Witt Exercise,readingnewspaper,watching 115Students
31 Exercise 1 1 Frequency 1 Regression Y Y N
2005) TV (USA)
179studentswho
(Gefen2003)
32 (Gefen2003) B2Cwebsite IS 1 1 TAM+HAB previously 1 CFA,SEM Y
(Developed)
purchasedonline
160inhabitantsof
(Klckner,Matthies Multiple
33 TravelModeChoice 1 1 RFM Bochum, 1 Y
&Hunecke2003) regression
Germany
31,144and94
(Limayem,M.&Hirt OReillysWebBoard3.5(Education (Limayem,Hirt& undergraduate
34 IS 1 1 TPB 1 PLS Y
2003) tool) Cheung2003) andmaster
students
(Limayem,M.,
TPB+IS (Limayem,Hirt& 1371,495and271
35 Cheung&Chan BlackboardLearningSystem IS 1 1 1 PLS y
continuance Cheung2003) students
2003)
CFA,PLS
(Limayem,Hirt& 227 convergent
(Limayem,M.,Hirt&
36 WWW IS 1 1 IScontinuance Cheung2003) undergraduate 1 validityand Y
Cheung2003)
(Developed) students discriminant
validity
RFM(Verplanken,B,
Aarts,H.,van
93,86,143,and76
Knippenberg,A.,&
(Verplanken& students(The Reliabilityand
37 Travelmodechoice Transport 1 1 vanKnippenberg,C. 1
Orbell2003) Netherlands, validity
1994)andSRHI
Norway)
(Verplanken,Bas&
Orbell2003)
RFM(Verplanken,B,
TPB,TIBTheory
Aarts,H.,van
(Bamberg&Schmidt of
38 CarUse Transport 1 Knippenberg,A.,& 608students 1 SEM Y
2003) Interpersonal
vanKnippenberg,C.
behavior
1994)
84ecstasyusers
(Orbelletal.2001)
39 (Orbelletal.2001) Ecstasyuse Health 1 1 TPB (Snowballing 1 Regression Y Y
(Developed)
technique)
(Towler&Shepherd
860householders
(Saba,Vassallo& Foodordrinks 1992;Tuorila&
40 Consumptionoffatcontainingfoods 1 1 TRA randomly 1 Regression Y Y
Turrini2000) consumption Pangborn1988)
selected(Italy)
(Adapted)
48and81sexually
TRA,TPB, (Trafimow2000) active Multiple
41 (Trafimow2000) Useofcondom Health 1 1 1 Y
TriandisTheory (Developed) undergraduates regression
USA
Frequency,habit
strenght,RFM
(Verplanken&Aarts Foodordrinks Multiple
42 Unhealthyfoodhabits 1 1 TPB (Verplanken,B.& 102students 1 Y
1999) consumption regression
Aarts1999)
(Developed)
(Ouellette&Wood 60research Bivariate
43 HabitandIntentioninEverydayLife Various 1 1 Frequency 1 Y
1998) reports Correlation
(Towler&Shepherd
909householders
(Saba&DiNatale Foodordrinks 1992;Tuorila&
44 Oliveoil 1 1 TRA randomly 1 SEM Y
1998a) consumption Pangborn1988)
selected(Italy)
(Adapted)
(Towler&Shepherd
929householders
(Saba&DiNatale Foodordrinks 1992;Tuorila&
45 Meatconsumption 1 1 TRA randomly 1 SEM Y Y
1998b) consumption Pangborn1988)
selected(Italy)
(Adapted)

(Source:Author)

227

Food&Drinks

Regression
Transport
Exercise
Health

Other

Other
SEM
PLS
ICT
No. Reference Behaviorofthestudy Theory Measurement SampleSize AnalysisType BI AB M

(Towler&Shepherd
111volunteersin
Foodordrinks 1992;Tuorila& Multiple
46 (Sabaetal.1998) Milkconsumption 1 1 TRA supermarkets 1 Y
consumption Pangborn1988) regression
(Rome,Italy)
(Adapted)
200+25randomly
(Verplankenetal. selected
47 TravelModeChoice Transport 1 1 TPB Frequency,RFM 1 Regression Y Y Y
1998) inhabitants(The
Netherlands)
38executive
(Bergeronetal. Correlation,
48 Executiveinformationsystems IS 1 1 Lengthofexperience information 1 Y
1995) regression
systemsusers
Hierarchical
49 Verplanken1994) CarUse Transport 1 RFM(original) 258adults 1 multiple Y
regression
288Recruitsata
(Towler&Shepherd Foodordrinks Frequncyandhabit Correlation,
50 Consumptionofahighfatfood(chips) 1 1 TRA countryshow 1 Y Y
1992) consumption selfreport regression
(UK)
16Research
51 (Ajzen1991) Diversebehaviors(Aggregateprinciple) Various 1 1 TPB Pastbehavior 1 Regression Y
reports
Correlation,
(Montano&Taplin PastBehavior 946womenage
52 Mammographyparticipation Health 1 1 TRA 1 Multiple Y N
1991) (Previoususe) 40andabove
regression
240 Multiple
(Bagozzi&Warshaw
53 Losingweight Health 1 1 Frequency undergraduate 1 regression,and Y N
1990)
students(Canada) logit
(Charng,Piliavin& TRAand 658blooddonors
54 Blooddonation Health 1 1 Frequency 1 Regression Y
Callero1988) identitytheory (USA)
(Mittal1988) 197adultrandom
55 (Mittal1988) SeatBeltUsage Transport 1 1 1 Regression Y Y
(Developed) sample(USA)
(Wittenbraker, Multiple
56 UseofSeatBelt Transport 1 1 TRA Frequency 134students 1 Y
Gibbs&Kahle1983) Regression
157students,
57 (Bagozzi1981) Blooddonation Health 1 1 Pastbehavior 1 Regression Y
facultyandstaff
Classroom
(Landis,Triandis& 77School a
58 Classroomteacherbehavior teacher 1 1 Frequency(Observed) 1 Regression Y
Adamopoulos1978) teachers
behavior

(Source:Author)

228

 APPENDIX6QSORTINGEXERCISE:LISTOFITEMS

Candidatesforelimination

EliminatedforReliability

Usedinmodeltests
basedonQSorting

EliminatedinEFA
Reverseditem

Measurementitemsasusedinthisthesis
ItemId

(7pointLikertScale) Sourcesoftheitem

AB1 Y Inthelastmonth,IspentalotoftimeusingiPhone. (Liangetal.2010)

AB2 Y Inthelastmonth,IusediPhonefrequently. (Liangetal.2010)
AB3 Y Inthelastmonth,IusediPhoneintensively. (Liangetal.2010)
BI1 Y IpredictIwoulduseiPhoneinthenext4weeks. (Venkateshetal.2003)
BI2 Y IintendtouseiPhoneinthenextmonth. (Venkateshetal.2003)
BI3 Y IplantouseiPhoneinthenext30days. (Venkateshetal.2003)
EE1 Y MyinteractionwithiPhonehasbeenclearandunderstandable. (Venkateshetal.2003)
EE2 Y IthasbeeneasyformetobecomeskillfulusingiPhone. (Venkateshetal.2003)
EE3 Y IhavefoundiPhoneeasytouse. (Venkateshetal.2003)
EE4 Y LearningtooperateiPhonehasbeeneasyforme. (Venkateshetal.2003)
FC1 Y IhavetheresourcesnecessarytouseiPhone. (Venkateshetal.2003)
FC2 Y IhavetheknowledgenecessarytouseiPhone. (Venkateshetal.2003)
FC3 Y Y Y iPhoneisnotcompatiblewithothersystemsIuse. (Venkateshetal.2003)
Aspecificperson(orgroup)isavailableforassistancewithiPhone
FC4 Y Y (Venkateshetal.2003)
difficulties.
Idon'tneedtothinkmuchonhowtouseiPhoneaseverythingI
HTF1 Y (Source:Author)
havebeendoinginmylifeissoclosetoit.
HTF2 Y UsingiPhonefitsverywellwithmyhabits. (Source:Author)
UsingiPhonefrequently,requiresmetochangemyhabitsinan
HTF3 Y Y Y (Source:Author)
uncomfortableway.
IncludingtheuseofiPhoneinmylifeiscompatiblewithmynormal
HTF4 Y (Source:Author)
behavior.
HTF5 Y IthinkusingiPhonedoesn'tsetmeapartfrommyhabits. (Source:Author)
Itendtousetechnologieswhichseemsomehowverysimilarto
HTF6 Y Y (Source:Author)
iPhone.
WorkingorplayingwithiPhonegoesverywellwiththewaysIhave
HTF7 Y (Source:Author)
learnthowtodothings.
UsingiPhoneimprovesmyperformanceinthecontextwhereIuse
PE1 Y (Venkateshetal.2003)
it.
PE2 Y IfindiPhoneusefulinmyjob. (Venkateshetal.2003)
PE3 Y UsingiPhoneenablesmetoaccomplishtasksmorequickly. (Venkateshetal.2003)
PE4 Y UsingiPhoneincreasesmyproductivity. (Venkateshetal.2003)
PE5 Y IfIuseiPhone,Iwillincreasemychancesofgettingaraise. (Venkateshetal.2003)
PE5b Y IfIusediPhone,Iwouldincreasemychancesofgettingrewarded. Adapted(Venkateshetal.2003)
PE6 IfindiPhoneusefulinatleastonethingthatIwantorneedtodo. Adapted(Venkateshetal.2003)
UsingiPhonecanenablemetoaccomplishwhatIwantorneedtodo
PE7 Adapted(Venkateshetal.2003)
inabetterway.
PE8 UsingiPhoneincreasesmycapacityfordoingwhatIwantorneed. Adapted(Venkateshetal.2003)
SI1 Y PeoplewhoareimportanttomethinkthatIshoulduseiPhone. (Venkateshetal.2003)
SI2 Y PeoplewhoinfluencemybehaviorthinkthatIshoulduseiPhone. (Venkateshetal.2003)
Peoplewithsomekindofauthorityinmylife,havebeenhelpfulin
SI3 Y (Venkateshetal.2003)
theuseofiPhone.
SI4 Y Y Ingeneral,myenvironmenthassupportedtheuseofiPhone. (Venkateshetal.2003)

229

 OPENEXERCISENAMESGIVENTOTHECATEGORIES
Code Examplesofnamesgivenbytherespondentstoits
Construct
Prefix items
ActualBehavior AB actualusage,actualuse,frequency,technologyusage,
aims, behavioral intentions, intention of usage,
BehavioralIntention BI
intentiontouse
behavioral familiarity/skills, capability and capacity for
using, capability and knowledge to use, ease of
EffortExpectancy EE adoption, ease of use, easiness, easy and simple
technology, easy use adaptation/learning new
technology,easy,intuitiveandeasytouse,lifestyle
adaptation/learning new technology, assistance,
capability and capacity for using, capability and
knowledge to use, consumer perception, compatibility,
FacilitatingConditions FC
difficulties, functionality, perceived ability, personal
factors, personal, usability, support system, system
compatibility,training
behavioral familiarity/skills, compatibility with
HabitTechnologyFit HTF lifestyle/habits, consumer habit, consumer perception,
environmentfit,habit,habits,lifestyle
advantage and disadvantage of use, advantage of
technology, advantage, advantages, an increase in my
performance,belief,benefitsoftechnologytomywork,
PerformanceExpectancy PE
benefits of using, benefits, benefits of technology,
consumer perception, increased productivity,
performance
influence of friends influence of others, influences,
SocialInfluence SI people and marketing influences, people's influence
overme,personinfluence,socialinfluence
OpenExerciseNamesGiventotheCategories (Source:Author)

230

 APPENDIX7PARALLELANALYSIS

RunMATRIXprocedure:
PARALLELANALYSIS:
PAF/CommonFactorAnalysis&
RawDataPermutation
SpecificationsforthisRun:

RawDataEigenvalues,&Mean&PercentileRandomDataEigenvalues
Root RawData Means Percentile
1.000 9.359964 427496 0.495914

2.000 2.494624 358470 0.410756

3.000 1.79586 307228 0.35167

4.000 1.162539 262699 0.303457
5.000 0.624308 222076 0.259737
6.000 0.449914 185416 0.219176
7.000 0.071162 151348 0.181688
8.000 0.052336 118352 0.147974
9.000 0.015501 85371 0.111895
10.000 0.007318 54227 0.080825
11.000 0.026137 24646 0.050455
12.000 0.028144 4544 0.021175
13.000 0.037519 33423 0.00874
14.000 0.049758 62452 0.03949
15.000 0.065365 89656 0.067895
16.000 0.06849 118305 0.093409
17.000 0.077895 147016 0.123245 Ncases 251
18.000 0.085961 176054 0.151036 Nvars 21
19.000 0.092162 206692 0.181717
Ndatsets 1000
20.000 0.119229 240191 0.212228
21.000 0.133405 281026 0.248081 Percent 95

ComputationaloptionsEFA

FACTOR
/VARIABLES AB1 AB2 AB3 BI1 BI2 BI3 EE1 EE2 EE3 EE4 FC1 FC2 FC3 FC4 HTF1 HTF2 HTF3 HTF4 HTF5
HTF7 PE1 PE2 PE3 PE4 SI1 SI2 SI3 SI4
/MISSING LISTWISE
/ANALYSIS AB1 AB2 AB3 BI1 BI2 BI3 EE1 EE2 EE3 EE4 FC1 FC2 FC3 FC4 HTF1 HTF2 HTF3 HTF4 HTF5
HTF7 PE1 PE2 PE3 PE4 SI1 SI2 SI3 SI4
/PRINT UNIVARIATE INITIAL CORRELATION DET KMO EXTRACTION ROTATION
/FORMAT SORT BLANK(.35)
/PLOT EIGEN ROTATION
/CRITERIA FACTORS(7) ITERATE(25)
/EXTRACTION PAF
/CRITERIA ITERATE(25)
/ROTATION VARIMAX
/METHOD=
CORRELATION.

(Source:Author)

231

 APPENDIX8SPSSSYNTAX:RELIABILITY,EFA,ANDPARALLELANALYSIS

Adaptedfrom:(Oconnor2000)

1RELIABILITY 41/METHOD=CORRELATION. 68 whethernormallydistributedrandom

2/VARIABLES=PE1PE2PE34bs 42 datagenerationor
3/SCALE('ALLVARIABLES')ALL 43 69permutationsoftherawdatasetareto
4/MODEL=ALPHA 44 beusedinthe
5/STATISTICS=DESCRIPTIVE 45 * Parallel Analysis Program For Raw Data 70parallelanalyses.
6/SUMMARY=TOTAL. andDataPermutations. 71
7 46 72*Permutationsoftherawdatasetcanbe
8RELIABILITY 47 * To run this program you need to first timeconsuming;
9/VARIABLES=E1E234bs specifythedata 73 Each parallel data set is based on
10/SCALE('ALLVARIABLES')ALL 48 for analysis and then RUN, all at once, columnwiserandomshufflings
11/MODEL=ALPHA thecommands 74ofthevaluesintherawdatamatrixusing
12/STATISTICS=DESCRIPTIVE 49fromtheMATRIXstatementtotheEND Castellan's(1992,
13/SUMMARY=TOTAL. MATRIXstatement. 75 BRMIC, 24, 7277) algorithm; The
14 50 distributionsoftheoriginal
15RELIABILITY 51*Thisprogramconductsparallelanalyses 76rawvariablesareexactlypreservedinthe
16/VARIABLES=SI1SI23bs ondatafilesinwhich shuffledversionsused
17/SCALE('ALLVARIABLES')ALL 52 the rows of the data matrix are 77intheparallelanalyses;Permutationsof
18/MODEL=ALPHA cases/individualsandthe therawdatasetare
19/STATISTICS=DESCRIPTIVE 53 columns are variables; Data are 78 thus highly accurate and most relevant,
20/SUMMARY=TOTAL. read/enteredintotheprogram especiallyincaseswhere
21 54 using the GET command (see the GET 79therawdataarenotnormallydistributed
22RELIABILITY commandbelow);TheGET orwhentheydonotmeet
23/VARIABLES=HTF24bsHTF5HTF7 55commandreadsanSPSSdatafile,which 80theassumptionofmultivariatenormality
24/SCALE('ALLVARIABLES')ALL canbeeitherthe (seeLongman&Holden,
25/MODEL=ALPHA 56 current, active SPSS data file or a 81 1992, BRMIC, 24, 493, for a Fortran
26/STATISTICS=DESCRIPTIVE previouslysaveddatafile; version);Ifyouwould
27/SUMMARY=TOTAL. 57 A valid filename/location must be 82liketogothisroute,itisperhapsbestto
28 specifiedontheGETcommand; (1)firstruna
29 58Asubsetofvariablesfortheanalysescan 83 normally distributed random data
30FACTOR bespecifiedbyusing generationparallelanalysisto
31/VARIABLESHTF24bsHTF5HTF7PE1PE2 59the"/VAR="subcommandwiththeGET 84familiarizeyourselfwiththeprogramand
PE34bsE1E234bsSI1SI23bs statement;Therecanbe togetaballpark
32/MISSINGLISTWISE 60nomissingvalues. 85 reference point for the number of
33 /ANALYSIS HTF24bs HTF5 HTF7 PE1 PE2 61 factors/components;
PE34bsE1E234bsSI1SI23bs 62*Youmustalsospecify: 86 (2) then run a permutations of the raw
34 /PRINT UNIVARIATE INITIAL 63 the # of parallel data sets for the dataparallelanalysis
CORRELATION DET KMO EXTRACTION analyses; 87 using a small number of datasets (e.g.,
ROTATION 64 the desired percentile of the 100),justtoseehowlong
35/FORMATSORTBLANK(.35) distributionandrandom 88theprogramtakestorun;then(3)runa
36/PLOTROTATION 65dataeigenvalues; permutationsoftheraw
37/CRITERIAFACTORS(8)ITERATE(25) 66 whether principal components 89 data parallel analysis using the number
38/EXTRACTIONPAF analysesorprincipalaxis/common ofparalleldatasetsthat
39/CRITERIAITERATE(25) 67 factor analysis are to be conducted, 90 you would like use for your final
40/ROTATIONVARIMAX and analyses;1000datasetsare

232

91 usually sufficient, although more 133computenvars=ncol(raw). 181callsetdiag(cr,smc).

datasetsshouldbeusedif 134 182computerealeval=eval(cr).
92thereareclosecalls. 135 * principal components analysis & 183 compute evals = make(nvars,ndatsets,
93 randomnormaldatageneration. 9999).
94 136doif(kind=1andrandtype=1). 184computenm1=1/(ncases1).
95*Thesenextcommandsgenerateartificial 137computenm1=1/(ncases1). 185loop#nds=1tondatsets.
rawdata 138 compute vcv = nm1 * (sscp(raw) 186 compute x = sqrt(2 *
96 (500 cases) that can be used for a trial ((t(csum(raw))*csum(raw))/ncases)). (ln(uniform(ncases,nvars))*1))&*
runof 139computed=inv(mdiag(sqrt(diag(vcv)))). 187 cos(6.283185 *
97theprogram,insteadofusingyourown 140computerealeval=eval(d*vcv*d). uniform(ncases,nvars)).
rawdata; 141 compute evals = make(nvars,ndatsets, 188 compute vcv = nm1 * (sscp(x)
98 Just select and run this whole file; 9999). ((t(csum(x))*csum(x))/ncases)).
However,makesureto 142loop#nds=1tondatsets. 189computed=inv(mdiag(sqrt(diag(vcv)))).
99 delete the artificial data commands 143 compute x = sqrt(2 * 190computer=d*vcv*d.
beforeattemptingto (ln(uniform(ncases,nvars))*1))&* 191computesmc=1(1&/diag(inv(r))).
100runyourowndata. 144 cos(6.283185 * 192callsetdiag(r,smc).
101 uniform(ncases,nvars)). 193computeevals(:,#nds)=eval(r).
102 set mxloops=9000 printback=off 145 compute vcv = nm1 * (sscp(x) 194endloop.
width=80seed=1953125.103matrix. ((t(csum(x))*csum(x))/ncases)). 195endif.
104 146computed=inv(mdiag(sqrt(diag(vcv)))). 196
105 * Enter the name/location of the data 147computeevals(:,#nds)=eval(d*vcv*d). 197 * PAF/common factor analysis & raw
fileforanalysesafter"FILE="; 148endloop. datapermutation.
106 If you specify "FILE = *", then the 149endif. 198doif(kind=2andrandtype=2).
programwillreadthecurrent, 150 199computenm1=1/(ncases1).
107 active SPSS data file; Alternatively, 151 * principal components analysis & raw 200 compute vcv = nm1 * (sscp(raw)
enterthename/location datapermutation. ((t(csum(raw))*csum(raw))/ncases)).
108 of a previously saved SPSS data file 152doif(kind=1andrandtype=2). 201computed=inv(mdiag(sqrt(diag(vcv)))).
insteadof"*"; 153computenm1=1/(ncases1). 202computecr=(d*vcv*d).
109youcanusethe"/VAR="subcommand 154 compute vcv = nm1 * (sscp(raw) 203computesmc=1(1&/diag(inv(cr))).
after"/missing=omit" ((t(csum(raw))*csum(raw))/ncases)). 204callsetdiag(cr,smc).
110subcommandtoselectvariablesforthe 155computed=inv(mdiag(sqrt(diag(vcv)))). 205computerealeval=eval(cr).
analyses. 156computerealeval=eval(d*vcv*d). 206 compute evals = make(nvars,ndatsets,
111GETraw/FILE=*/missing=omit/VAR= 157 compute evals = make(nvars,ndatsets, 9999).
HTF24bs HTF5 HTF7 PE1 PE2 PE34bs E1 9999). 207computenm1=1/(ncases1).
E234bsSI1SI23bs. 158loop#nds=1tondatsets. 208loop#nds=1tondatsets.
112 159computex=raw. 209computex=raw.
113 * Enter the desired number of parallel 160loop#c=1tonvars. 210loop#c=1tonvars.
datasetshere. 161loop#r=1to(ncases1). 211loop#r=1to(ncases1).
114computendatsets=1000. 162 compute k = trunc( (ncases #r + 1) * 212 compute k = trunc( (ncases #r + 1) *
115 uniform(1,1)+1)+#r1. uniform(1,1)+1)+#r1.
116*Enterthedesiredpercentilehere. 163computed=x(#r,#c). 213computed=x(#r,#c).
117computepercent=95. 164computex(#r,#c)=x(k,#c). 214computex(#r,#c)=x(k,#c).
118 165computex(k,#c)=d. 215computex(k,#c)=d.
119*Entereither 166endloop. 216endloop.
1201forprincipalcomponentsanalysis,or 167endloop. 217endloop.
121 2 for principal axis/common factor 168 compute vcv = nm1 * (sscp(x) 218 compute vcv = nm1 * (sscp(x)
analysis. ((t(csum(x))*csum(x))/ncases)). ((t(csum(x))*csum(x))/ncases)).
122computekind=2. 169computed=inv(mdiag(sqrt(diag(vcv)))). 219computed=inv(mdiag(sqrt(diag(vcv)))).
123 170computeevals(:,#nds)=eval(d*vcv*d). 220computer=d*vcv*d.
124*Entereither 171endloop. 221computesmc=1(1&/diag(inv(r))).
1251fornormallydistributedrandomdata 172endif. 222callsetdiag(r,smc).
generationparallelanalysis,or 173 223computeevals(:,#nds)=eval(r).
1262forpermutationsoftherawdataset. 174*PAF/commonfactoranalysis&random 224endloop.
127computerandtype=2. normaldatageneration. 225endif.
128 175doif(kind=2andrandtype=1). 226
129 176computenm1=1/(ncases1). 227 * identifying the eigenvalues
130 ****************** End of user 177 compute vcv = nm1 * (sscp(raw) correspondingtothedesiredpercentile.
specifications.****************** ((t(csum(raw))*csum(raw))/ncases)). 228 compute num =
131 178computed=inv(mdiag(sqrt(diag(vcv)))). rnd((percent*ndatsets)/100).
132computencases=nrow(raw). 179computecr=(d*vcv*d). 229 compute results = { t(1:nvars), realeval,
180computesmc=1(1&/diag(inv(cr))). t(1:nvars),t(1:nvars)}.

233

230loop#root=1tonvars. 266 print /title="for the same roots. The 296endmatrix.

231computeranks=rnkorder(evals(#root,:)).
232loop#col=1tondatsets.
233doif(ranks(1,#col)=num).
234 compute results(#root,4)
evals(#root,#col).
235break.
236endif.
237endloop.
238endloop.
239 compute results(:,3) = rsum(evals) /
ndatsets.
240
241print/title="PARALLELANALYSIS:".
242doif(kind=1andrandtype=1).
243 print /title="Principal Components &
RandomNormalDataGeneration".
244elseif(kind=1andrandtype=2).
245 print /title="Principal Components &
RawDataPermutation".
246elseif(kind=2andrandtype=1).
247 print /title="PAF/Common Factor
Analysis & Random Normal Data
Generation".
248elseif(kind=2andrandtype=2).
249 print /title="PAF/Common Factor
Analysis&RawDataPermutation".
250endif.
251 compute specifs = {ncases; nvars;
ndatsets;percent}.
252printspecifs/title="Specificationsforthis
Run:"
253 /rlabels="Ncases" "Nvars" "Ndatsets"
"Percent".
254printresults
255/title="RawDataEigenvalues,&Mean&
PercentileRandomDataEigenvalues"
256 /clabels="Root" "Raw Data" "Means"
"Prcntyle"/format"f12.6".
257
258doif(kind=2).
259print/space=1.
260print/title="Warning:Parallelanalysesof
adjustedcorrelationmatrices".
261 print /title="eg, with SMCs on the
diagonal,tendtoindicatemorefactors".
262 print /title="than warranted (Buja, A., &
Eyuboglu, N., 1992, Remarks on
parallel".
263 print /title="analysis. Multivariate
BehavioralResearch,27,509540.).".
264 print /title="The eigenvalues for trivial,
negligiblefactorsinthereal".
265 print /title="data commonly surpass
corresponding random
eigenvalues".
eigenvaluesfromparallelanalyses". 297
267 print /title="can be used to determine 298 * plots the eigenvalues, by root, for the
therealdataeigenvaluesthatare". real/rawdataandfortherandomdata;
= 268 print /title="beyond chance, but 299 This command works in SPSS 12, but
additional procedures should then be notinallearlierversions.
used". 300GETfile='screedata.sav'.
269print/title="totrimtrivialfactors.". 301 TSPLOT VARIABLES= rawdata means
270print/space=2. percntyl/ID=root/NOLOG.
271 print /title="Principal components
eigenvalues are often used to
determine".
272 print /title="the number of common
factors.Thisisthedefaultinmost".
273 print /title="statistical software
packages,anditistheprimarypractice".
274print/title="intheliterature.Itisalsothe
methodusedbymanyfactor".
275 print /title="analysis experts, including
Cattell,whooftenexamined".
276 print /title="principal components
eigenvalues in his scree plots to
determine".
277 print /title="the number of common
factors. But others believe this
common".
278 print /title="practice is wrong. Principal
componentseigenvaluesarebased".
279 print /title="on all of the variance in
correlationmatrices,includingboth".
280 print /title="the variance that is shared
amongvariablesandthevariances".
281 print /title="that are unique to the
variables.Incontrast,principal".
282 print /title="axis eigenvalues are based
solelyonthesharedvariance".
283 print /title="among the variables. The
twoproceduresarequalitatively".
284 print /title="different. Some therefore
claimthattheeigenvaluesfromone".
285 print /title="extraction method should
notbeusedtodetermine".
286 print /title="the number of factors for
theotherextractionmethod.".
287print/title="Theissueremainsneglected
andunsettled.".
288endif.
289
290computeroot=results(:,1).
291computerawdata=results(:,2).
292computepercntyl=results(:,4).
293
294 save results /outfile= 'screedata.sav' /
var=rootrawdatameanspercntyl.
data 295

234

 APPENDIX9CORRELATIONMATRIX

Correlations
PE EE SI HTF BI AB GEN AGE_1 EXP_1 VOL1 GENxPE GENxEE GENxSI GENxHTF AGExPE AGExEE AGExSI AGExHTF EXPxEE EXPxSI VOLxSI EXPxHTF
Pearson Correlation 1 .439 **
.459 **
.664 **
.477 **
.287 **
.028 .107 *
.177 **
.237 **
.949** .414** .431** .621** .969** .426** .441** .639** .364** .338** .427** .554**
PE Sig. (2-tailed) .000 .000 .000 .000 .000 .535 .016 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .439** 1 .317 **
.644 **
.418 **
.328 **
.034 .086 .315 **
.426 **
.415 **
.950 **
.286 **
.608 **
.419 **
.973 **
.313 **
.614 **
.819 **
.241 **
.261 **
.519**
EE Sig. (2-tailed) .000 .000 .000 .000 .000 .444 .053 .000 .000 .000 .000 .000 .000 .000 0.000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .459** .317** 1 .382 **
.517 **
.394 **
-.007 .051 .267 **
.224 **
.431 **
.285 **
.949 **
.353 **
.443 **
.320 **
.972 **
.370 **
.297 **
.803 **
.952 **
.349**
SI Sig. (2-tailed) .000 .000 .000 .000 .000 .877 .249 .000 .000 .000 .000 .000 .000 .000 .000 0.000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .664** .644** .382** 1 .557** .394** -.017 .047 .271** .394** .619** .604** .351** .949** .630** .616** .363** .971** .526** .288** .339** .809**
HTF Sig. (2-tailed) .000 .000 .000 .000 .000 .704 .296 .000 .000 .000 .000 .000 .000 .000 .000 .000 0.000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .477** .418** .517** .557** 1 .696** -.037 .043 .542** .314** .469** .391** .476** .530** .459** .402** .489** .552** .480** .416** .480** .590**
BI Sig. (2-tailed) .000 .000 .000 .000 .000 .412 .332 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .287** .328** .394** .394** .696** 1 .001 .000 .653** .140** .282** .306** .361** .384** .273** .322** .374** .390** .499** .460** .378** .580**
AB Sig. (2-tailed) .000 .000 .000 .000 .000 .979 .992 .000 .002 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .028 .034 -.007 -.017 -.037 .001 1 -.079 .051 -.056 .025 .031 -.006 -.016 .025 .022 -.017 -.033 .041 -.013 .000 .027
GEN Sig. (2-tailed) .535 .444 .877 .704 .412 .979 .077 .258 .210 .570 .485 .887 .727 .581 .623 .696 .463 .358 .777 .998 .542
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .107* .086 .051 .047 .043 .000 -.079 1 -.057 .062 .101* .068 .035 .024 .110* .091* .041 .032 .052 .003 .037 .015
AGE_1 Sig. (2-tailed) .016 .053 .249 .296 .332 .992 .077 .201 .165 .023 .127 .428 .589 .013 .040 .359 .480 .243 .946 .403 .741
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .177** .315** .267** .271** .542** .653** .051 -.057 1 .163** .194** .311** .255** .285** .161** .302** .243** .257** .574** .392** .239** .498**
EXP_1 Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .258 .201 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .237** .426** .224** .394** .314** .140** -.056 .062 .163** 1 .228** .414** .224** .377** .218** .405** .209** .363** .333** .163** .125** .281**
VOL1 Sig. (2-tailed) .000 .000 .000 .000 .000 .002 .210 .165 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .005 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .949** .415** .431** .619** .469** .282** .025 .101* .194** .228** 1 .435** .451** .645** .913** .398** .410** .595** .365** .321** .398** .535**
GENxPE Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .570 .023 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .414 **
.950 **
.285 **
.604 **
.391 **
.306 **
.031 .068 .311 **
.414 **
.435 **
1 .287 **
.634 **
.390 **
.919 **
.281 **
.573 **
.796 **
.228 **
.230 **
.496**
GENxEE Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .485 .127 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .431** .286** .949** .351** .476** .361** -.006 .035 .255 **
.224 **
.451 **
.287 **
1 .362 **
.412 **
.289 **
.915 **
.342 **
.284 **
.783 **
.901 **
.336**
GENxSI Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .887 .428 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .621** .608** .353** .949** .530** .384** -.016 .024 .285 **
.377 **
.645 **
.634 **
.362 **
1 .589 **
.578 **
.336 **
.919 **
.509 **
.278 **
.311 **
.780**
GENxHTF Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .727 .589 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .969** .419** .443** .630** .459** .273** .025 .110 *
.161 **
.218 **
.913 **
.390 **
.412 **
.589 **
1 .439 **
.453 **
.649 **
.344 **
.321 **
.417 **
.523**
AGExPE Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .581 .013 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .426** .973** .320** .616** .402** .322** .022 .091* .302** .405** .398** .919** .289** .578** .439** 1 .340** .624** .795** .239** .271** .493**
AGExEE Sig. (2-tailed) .000 0.000 .000 .000 .000 .000 .623 .040 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .441** .313** .972** .363** .489** .374** -.017 .041 .243** .209** .410** .281** .915** .336** .453** .340** 1 .375** .289** .763** .931** .324**
AGExSI Sig. (2-tailed) .000 .000 0.000 .000 .000 .000 .696 .359 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .639** .614** .370** .971** .552** .390** -.033 .032 .257** .363** .595** .573** .342** .919** .649** .624** .375** 1 .501** .272** .334** .780**
AGExHTF Sig. (2-tailed) .000 .000 .000 0.000 .000 .000 .463 .480 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .364** .819** .297** .526** .480** .499** .041 .052 .574** .333** .365** .796** .284** .509** .344** .795** .289** .501** 1 .312** .276** .640**
EXPxEE Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .358 .243 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .338** .241** .803** .288** .416** .460** -.013 .003 .392** .163** .321** .228** .783** .278** .321** .239** .763** .272** .312** 1 .782** .395**
EXPxSI Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .777 .946 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .427** .261** .952** .339** .480** .378** .000 .037 .239** .125** .398** .230** .901** .311** .417** .271** .931** .334** .276** .782** 1 .337**
VOLxSI Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .998 .403 .000 .005 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
Pearson Correlation .554** .519** .349** .809** .590** .580** .027 .015 .498** .281** .535** .496** .336** .780** .523** .493** .324** .780** .640** .395** .337** 1
EXPxHTF Sig. (2-tailed) .000 .000 .000 .000 .000 .000 .542 .741 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
N 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503 503
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).

(Source:Author)

235

 APPENDIX10BASE,EXTENDEDANDMODIFIEDMODELSWITHSMARTPLS

BaseModelwithandwithoutModerators:UTAUT
AnalysisSoftware:SmartPLS2.0
Purpose:Crossvalidationofthemainanalysistechnique

ModelfromVenkateshetal.(2003),Data&Analysissource:(Author)

236

Extendedmodelwithandwithoutmoderators:UTAUTbasemodel+HTF
AnalysisSoftware:SmartPLS2.0
Purpose:Crossvalidationofthemainanalysistechnique

BasemodelfromVenkateshetal.(2003),extension(HTF),Data&Analysissource:
(Author)

237

ModifiedModelwithandwithoutModerators:HTFModel
AnalysisSoftware:SmartPLS2.0
Purpose:Crossvalidationofthemainanalysistechnique

(Source:Author)

APPENDIX11UTAUTWITHPLSGRAPH

238

UTAUTBaseModelwithModerators
AnalysisSoftware:PLSGraph3.0
OriginalSoftwareusedbyVenkateshetal.(2003):PLSGraphVersion2.91.03.04
Purpose:Crossvalidationandanalysisreplication

ModelfromVenkateshetal.(2003),Data&Analysissource:(Author)

239

 APPENDIX12 ETHICSAPPROVAL

240

APPENDIX13UNIONSANDINTERCEPTSCALCULATIONS

MSExcelspreadsheetanditsformulas
BasedonByrne(2002)andPolkowski(2013)
A B C D
1 Id. Area R2 f2
2 1 HTF 0.316 =C2/(1C2)
3 2 PE 0.23 =C3/(1C3)
4 3 EE 0.185 =C4/(1C4)
5 4 SI 0.269 =C5/(1C5)
6 5 HTFPE 0.335 =C6/(1C6)
7 6 HTFEE 0.323 =C7/(1C7)
8 7 HTFSI 0.423 =C8/(1C8)
9 8 PEEE 0.287 =C9/(1C9)
10 9 PESI 0.342 =C10/(1C10)
11 10 EESI 0.345 =C11/(1C11)
12 11 HTFPEEE 0.341 =C12/(1C12)
13 12 HTFPESI 0.425 =C13/(1C13)
14 13 HTFEESI 0.425 =C14/(1C14)
15 14 PEEESI 0.378 =C15/(1C15)
16 15 HTFPEEESI 0.427 =C16/(1C16)
17 A HTF =C16C15 =C17/(1C17)
18 B PE =C16C14 =C18/(1C18)
19 C EE =C16C13 =C19/(1C19)
20 D SI =C16C12 =C20/(1C20)
21 E HTFPE =C16C11C18C17 =C21/(1C21)
22 F HTFEE =C16C10C19C17 =C22/(1C22)
23 G HTFSI =$C$16C9C17C20 =C23/(1C23)
24 H PEEE =$C$16C8C18C19 =C24/(1C24)
25 I PESI =$C$16C7C18C20 =C25/(1C25)
26 J EESI =$C$16C6C19C20 =C26/(1C26)
27 K HTFPEEE =$C$16C5C18C21C24C19C22C17 =C27/(1C27)
28 L HTFPESI =$C$16C4C18C21C25C20C23C17 =C28/(1C28)
29 M HTFEESI =$C$16C3C19C22C17C26C20C23 =C29/(1C29)
30 N PEEESI =$C$16C2C18C24C19C25C26C20 =C30/(1C30)
31 O HTFPEEESI =$C$16SUM(C17:C30) =C31/(1C31)
(Source:Author)

241