Professional Documents
Culture Documents
COMPLETE
BUSINESS
STATISTICS
by
AMIR D. ACZEL
&
JAYAVEL SOUNDERPANDIAN
6th edition (SIE)
9-2
Chapter 9
Analysis of Variance
9-3
9 Analysis of Variance
Using Statistics
The Hypothesis Test of Analysis of Variance
The Theory and Computations of ANOVA
The ANOVA Table and Examples
Further Analysis
Models, Factors, and Designs
Two-Way Analysis of Variance
Blocking Designs
9-4
9 LEARNING OBJECTIVES
After studying this chapter you should be able to:
9-5
In an analysis of variance:
We
9-6
9-7
arenormally
normallydistributed,
distributed,
are
withmeans
meansi ithat
thatmay
mayor
ormay
maynot
notbe
beequal,
equal,
with
butwith
withequal
equalvariances,
variances,i2i.2.
but
Population 1
Population 2
Population 3
Estimateofofvariance
variancebased
basedon
onmeans
meansfrom
fromr rsamples
samples
Estimate
Estimateofofvariance
variancebased
basedon
onall
allsample
sampleobservations
observations
Estimate
Thatis,
is,the
thetest
teststatistic
statisticininan
ananalysis
analysisof
ofvariance
varianceisisbased
basedon
onthe
theratio
ratioof
of
That
twoestimators
estimatorsof
ofaapopulation
populationvariance,
variance,and
andisistherefore
thereforebased
basedon
onthe
theFF
two
distribution,with
with(r-1)
(r-1)degrees
degreesof
offreedom
freedomininthe
thenumerator
numeratorand
and(n-r)
(n-r)
distribution,
degreesof
offreedom
freedomininthe
thedenominator.
denominator.
degrees
9-8
9-9
Wewould
wouldexpect
expectthe
thesample
samplemeans
meanstotobe
benearly
nearly
We
equal,as
asininthis
thisillustration.
illustration. And
Andwe
wewould
would
equal,
expectthe
thevariation
variationamong
amongthe
thesample
samplemeans
means
expect
(betweensample)
sample)totobe
besmall,
small,relative
relativetotothe
the
(between
variationfound
foundaround
aroundthe
theindividual
individualsample
sample
variation
means(within
(withinsample).
sample).
means
thenull
nullhypothesis
hypothesisisistrue,
true, the
thenumerator
numeratorinin
IfIfthe
thetest
teststatistic
statisticisisexpected
expectedtotobe
besmall,
small,relative
relative
the
thedenominator:
denominator:
totothe
=
(r-1, n-r)
FF(r-1,
n-r)=
x
9-10
Inany
anyof
ofthese
thesesituations,
situations,we
wewould
wouldnot
notexpect
expectthe
thesample
samplemeans
meanstotoall
allbe
benearly
nearly
In
equal. We
Wewould
wouldexpect
expectthe
thevariation
variationamong
amongthe
thesample
samplemeans
means(between
(between
equal.
sample)totobe
belarge,
large,relative
relativetotothe
thevariation
variationaround
aroundthe
theindividual
individualsample
samplemeans
means
sample)
(withinsample).
sample).
(within
thenull
nullhypothesis
hypothesisisisfalse,
false, the
thenumerator
numeratorininthe
thetest
teststatistic
statisticisisexpected
expectedtotobe
be
IfIfthe
large,relative
relativetotothe
thedenominator:
denominator:
large,
Estimate of variance based on means from r samples
= Estimate
of variance based on means from r samples
(r-1, n-r)
FF(r-1,
=
n-r)
9-11
drawan
anindependent
independentrandom
randomsample,
sample,with
withnn11++nn22++nn33++
draw
54. Then
Thenour
ourtest
teststatistic
statisticis:
is:
nn44==54.
Estimateofofvariance
variancebased
basedon
onmeans
meansfrom
from44samples
samples
= F(3,50) =Estimate
FF(4-1,
(4-1, 54-4)
54-4)= F(3,50) =
Estimateofofvariance
variancebased
basedon
onall
all54
54sample
sampleobservations
observations
Estimate
f(F)
0.5
0.4
0.3
0.2
=0.05
0.1
0.0
0
3
2.79
F(3,50)
9-12
Example 9-1
Randomlychosen
chosengroups
groupsofofcustomers
customerswere
wereserved
serveddifferent
differenttypes
typesofofcoffee
coffeeand
andasked
askedtotorate
ratethe
the
Randomly
coffeeon
onaascale
scaleofof00toto100:
100:21
21were
wereserved
servedpure
pureBrazilian
Braziliancoffee,
coffee,20
20were
wereserved
servedpure
pureColombian
Colombian
coffee
coffee,and
and22
22were
wereserved
servedpure
pureAfrican-grown
African-growncoffee.
coffee.
coffee,
Theresulting
resultingtest
teststatistic
statisticwas
wasFF==2.02
2.02
The
F Distribution with 2 and 60 Degrees of Freedom
F 2.02 F
2,60
3.15
0.7
0.6
0.5
f(F)
H0 : 1 2 3
H1: Not all three means equal
n1 = 21 n 2 = 20 n3 = 22 n = 21+ 20 + 22 = 63
r =3
The critical point for = 0.05 is:
F
F
F
3.15
r -1,n-r
31,633
2,60
0.4
0.3
0.2
=0.05
0.1
0.0
0
Test Statistic=2.02
F(2,60)=3.15
9-13
9-14
Sample point(j)
I = 1 Triangle
1
Triangle
2
Triangle
3
Triangle
4
Mean of Triangles
I = 2 Square
1
Square
2
Square
3
Square
4
Mean of Squares
I = 3 Circle
1
Circle
2
Circle
3
Mean of Circles
les
Grand mean of all data points
Value(x ij)
4
5
7
8
6
10
11
12
13
11.5
1
2
3
2
6.909
x1=6
x2=11.5
x=6.909
x3=2
0
10
9-15
xx
xx
ee
ij
ij
ij
ij
tt xx xx
i
TheANOVA
ANOVAprinciple
principlesays:
says:
The
Whenthe
thepopulation
populationmeans
meansare
arenot
notequal,
equal,the
theaverage
averageerror
error
When
(withinsample)
sample) isisrelatively
relativelysmall
smallcompared
comparedwith
withthe
theaverage
average
(within
treatment(between
(betweensample)
sample)deviation.
deviation.
treatment
9-16
Thetotal
totaldeviation
deviation(Tot
(Totij))isisthe
thedifference
differencebetween
betweenaadata
datapoint
point(x
(xij))and
andthe
thegrand
grandmean
mean(x):
(x):
The
ij
ij
Totij=x
=xij--xx
Tot
ij
ij
Forany
anydata
datapoint
pointxx:ij:
For
ij
Tot==tt++ee
Tot
Thatis:
is:
That
TotalDeviation
Deviation==Treatment
TreatmentDeviation
Deviation++Error
ErrorDeviation
Deviation
Total
Total deviation:
Tot24=x24-x=6.091
Error deviation:
e24=x24-x2=1.5
x24=13
Treatment deviation:
t2=x2-x=4.591
x2=11.5
x = 6.909
10
Squared Deviations
2
= (x x)
i
ij
i
2
2
Tot ij ( x ij x )
t
+e
( x ij x )
i
9-17
Tot
e
nt
+
i 1 j 1 ij
i 1 ii
i 1 j 1 ij
n
n
j
j
r
r
r
2
2
(x x) = n (x x)
( x x )2
i
i 1 j 1 ij
i 1 i i
i 1 j 1 ij
SST =
SSTR
SSE
TheSum
Sumof
ofSquares
SquaresPrinciple
Principle
The
Thetotal
totalsum
sumof
ofsquares
squares(SST)
(SST)isisthe
thesum
sumof
oftwo
twoterms:
terms: the
thesum
sumof
of
The
squaresfor
fortreatment
treatment(SSTR)
(SSTR)and
andthe
thesum
sumof
ofsquares
squaresfor
forerror
error(SSE).
(SSE).
squares
SST == SSTR
SSTR ++ SSE
SSE
SST
9-18
9-19
SSE
SST
SSTmeasures
measuresthe
thetotal
totalvariation
variationininthe
thedata
dataset,
set,the
thevariation
variationof
ofall
allindividual
individualdata
data
SST
pointsfrom
fromthe
thegrand
grandmean.
mean.
points
SSTRmeasures
measuresthe
theexplained
explainedvariation,
variation,the
thevariation
variationof
ofindividual
individualsample
samplemeans
means
SSTR
fromthe
thegrand
grandmean.
mean. ItItisisthat
thatpart
partofofthe
thevariation
variationthat
thatisispossibly
possiblyexpected,
expected,oror
from
explained,because
becausethe
thedata
datapoints
pointsare
aredrawn
drawnfrom
fromdifferent
differentpopulations.
populations. Its
Itsthe
the
explained,
variationbetween
betweengroups
groupsofofdata
datapoints.
points.
variation
SSEmeasures
measuresunexplained
unexplainedvariation,
variation,the
thevariation
variationwithin
withineach
eachgroup
groupthat
thatcannot
cannotbe
be
SSE
explainedby
bypossible
possibledifferences
differencesbetween
betweenthe
thegroups.
groups.
explained
9-20
9-21
Recallthat
thatthe
thecalculation
calculationofofthe
thesample
samplevariance
varianceinvolves
involvesthe
thedivision
divisionof
ofthe
thesum
sumofof
Recall
squareddeviations
deviationsfrom
fromthe
thesample
samplemean
meanby
bythe
thenumber
numberof
ofdegrees
degreesof
offreedom.
freedom. This
This
squared
principleisisapplied
appliedas
aswell
welltotofind
findthe
themean
meansquared
squareddeviations
deviationswithin
withinthe
theanalysis
analysisof
of
principle
variance.
variance.
Meansquare
squaretreatment
treatment(MSTR):
(MSTR):
Mean
SSTR
MSTR
(r 1)
Meansquare
squareerror
error(MSE):
(MSE):
Mean
SSE
MSE
(n r )
Meansquare
squaretotal
total(MST):
(MST):
Mean
SST
MST
(n 1)
(Notethat
thatthe
theadditive
additiveproperties
propertiesofofsums
sumsofofsquares
squaresdo
donot
notextend
extendtotothe
themean
mean
(Note
squares. MSTMSTR
MSTMSTR++MSE.
MSE.
squares.
9-22
n
(
)
2 when the null hypothesis is true
2
i
i
E ( MSTR)
r 1
> 2 when the null hypothesis is false
where i is the mean of population i and is the combined mean of all r populations.
That is, the expected mean square error (MSE) is simply the common population variance
(remember the assumption of equal population variances), but the expected treatment sum of
squares (MSTR) is the common population variance plus a term related to the variation of the
individual population means around the grand population mean.
If the null hypothesis is true so that the population means are all equal, the second term in
the E(MSTR) formulation is zero, and E(MSTR) is equal to the common population variance.
9-23
MSTR
MSE
9-24
9-25
(x ij -xi ) (x ij -x i )2
Value (x ij )
Triangle
-2
Triangle
-1
Triangle
Triangle
Square
10
-1.5
2.25
Square
Square
Square
2
2
2
2
3
4
11
12
13
-0.5
0.5
1.5
0.25
0.25
2.25
Circle
-1
Circle
Circle
17
73
Treatment
(x i -x)
(x i -x)
ni (x i -x)
Triangle
-0.909
0.826281
3.305124
Square
4.591
21.077281
84.309124
Circle
-4.909
124.098281
72.294843
159.909091
j
r
( x x ) 2 17
SSE
i
i 1 j 1 ij
r
2
SSTR n ( x x ) 159 .9
i 1 i i
SSTR
159 .9
79 .95
MSTR
r 1
( 3 1)
SSTR 17
2 .125
MSE
n r
8
MSTR
79 .95
37 .62 .
F
MSE
2 .125
( 2 ,8 )
Critical point ( = 0.01): 8.65
H may be rejected at the 0.01 level
0
of significance.
9-26
ANOVA Table
Source of
Variation
Sum of
Squares
Degrees of
Freedom Mean Square F Ratio
Treatment SSTR=159.9
(r-1)=2
MSTR=79.95 37.62
Error
SSE=17.0
(n-r)=8
MSE=2.125
Total
SST=176.9
(n-1)=10
MST=17.69
TheANOVA
ANOVATable
Tablesummarizes
summarizesthe
the
The
ANOVAcalculations.
calculations.
ANOVA
0.6
0.5
f(F)
0.4
0.3
0.2
0.01
0.1
0.0
0
8.65
10
F(2,8)
thisinstance,
instance,since
sincethe
thetest
teststatistic
statisticisis
InInthis
greaterthan
thanthe
thecritical
criticalpoint
pointfor
foran
an=
=
greater
0.01level
levelofofsignificance,
significance,the
thenull
null
0.01
hypothesismay
maybe
berejected,
rejected,and
andwe
wemay
may
hypothesis
concludethat
thatthe
themeans
meansfor
fortriangles,
triangles,
conclude
squares,and
andcircles
circlesare
arenot
notall
allequal.
equal.
squares,
9-27
Template Output
9-28
89
Source of
Variation
Martinique
75
Treatment
MSTR= 3552
Eleuthra
73
Error
SSE=98356
(n-r)= 195
MSE= 504.39
Paradise Island
91
Total
SST=112564
(n-1)= 199
MST= 565.65
St. Lucia
85
SSE=98356
Sum of
Squares
Degrees of
Freedom
Mean Square
F Ratio
7.04
0.4
f(F)
SST=112564
Mean Response (x i )
0.3
0.2
0.01
0.1
0.0
0
3.41
F(4,200)
Theresultant
resultantFF
The
ratioisislarger
largerthan
than
ratio
thecritical
criticalpoint
pointfor
for
the
=0.01,
0.01,so
sothe
the
=
nullhypothesis
hypothesismay
may
null
berejected.
rejected.
be
9-29
Sum of
Squares
Degrees of
Freedom
Mean Square
F Ratio
Treatment
SSTR= 879.3
(r-1)=3
MSTR= 293.1
8.52
Error
SSE= 18541.6
(n-r)= 539
MSE=34.4
Total
SST= 19420.9
(n-1)=542
MST= 35.83
Giventhe
thetotal
totalnumber
numberof
ofobservations
observations(n
(n==543),
543),the
thenumber
numberof
ofgroups
groups
Given
4),the
theMSE
MSE(34.
(34.4),
4),and
andthe
theFFratio
ratio(8.52),
(8.52),the
theremainder
remainderof
ofthe
theANOVA
ANOVA
(r(r==4),
tablecan
canbe
becompleted.
completed. The
Thecritical
criticalpoint
pointof
ofthe
theFFdistribution
distributionfor
for=
=0.01
0.01
table
and(3,
(3,400)
400)degrees
degreesof
offreedom
freedomisis3.83.
3.83. The
Thetest
teststatistic
statisticininthis
thisexample
exampleisis
and
muchlarger
largerthan
thanthis
thiscritical
criticalpoint,
point,so
sothe
theppvalue
valueassociated
associatedwith
withthis
thistest
test
much
statisticisisless
lessthan
than0.01,
0.01,and
andthe
thenull
nullhypothesis
hypothesismay
maybe
berejected.
rejected.
statistic
9-30
Do Not Reject H0
ANOVA
Stop
Reject H0
Further
Analysis
Confidence Intervals
for Population Means
Tukey Pairwise
Comparisons Test
Mean Response (x i )
Guadeloupe
89
Martinique
75
Eleuthra
73
Paradise Island
91
St. Lucia
85
SST = 112564
SSE = 98356
ni = 40
n = (5)(40) = 200
MSE = 504.39
MSE
504.39
xi 1.96
xi 6.96
ni
40
2
89 6.96 [82.04, 95.96]
75 6.96 [ 68.04,81.96]
73 6.96 [ 66.04, 79.96]
91 6.96 [84.04,97.96]
85 6.96 [ 78.04, 91.96]
xi t
9-31
9-32
T q
MSE
ni
r!
3:
9-33
H
:
1
Guadeloupe
89
H1:1 1 1 2 2
H1:1 2 2 4 4
Martinique
75
|89-75|=14>13.7*
|75-91|=16>13.7*
22
Martinique
75
|89-75|=14>13.7*
|75-91|=16>13.7*
Eleuthra
73
II. HH:0:13
VII. HH:0:25
33
Eleuthra
73
II.
VII.
0
1
3
0
2
5
4
Paradise
Is.
91
H
:
H
:
4
Paradise Is.
91
H1:1 1 1 3 3
H1:1 2 2 5 5
St.Lucia
Lucia
85
|89-73|=16>13.7*
|75-85|=10<13.7
55
St.
85
|89-73|=16>13.7*
|75-85|=10<13.7
III. HH:0:14
VIII.HH:0:34
III.
VIII.
0
1
4
0
3
4
The
critical
point
T
for
H
:
H
:
0.05for
The critical point T0.05
H1:1 1 1 4 4
H1:1 3 3 4 4
r=5and
and(n-r)=195
(n-r)=195
|89-91|=2<13.7
|73-91|=18>13.7*
r=5
|89-91|=2<13.7
|73-91|=18>13.7*
degreesofoffreedom
freedomis:
is:
IV. HH:0:15
IX. HH:0:35
degrees
IV.
IX.
0
1
5
0
3
5
MSE
H
:
H
:
T q
H1:1 1 1 5 5
H1:1 3 3 5 5
ni
|89-85|=4<13.7
|73-85|=12<13.7
|89-85|=4<13.7
|73-85|=12<13.7
504.4
V. HH:0:23
X. HH:0:45
V.
X.
0
2
3
0
4
5
3.86
13.7
H
:
H
:
40
H1:1 2 2 3 3
H1:1 4 4 5 5
|75-73|=2<13.7
|91-85|=6<13.7
6<13.7
|75-73|=2<13.7
|91-85|=
Rejectthe
thenull
nullhypothesis
hypothesisififthe
theabsolute
absolutevalue
valueofofthe
thedifference
differencebetween
betweenthe
thesample
samplemeans
means
Reject
greaterthan
thanthe
thecritical
criticalvalue
valueofofT.
T.(The
(Thehypotheses
hypothesesmarked
markedwith
with**are
arerejected.)
rejected.)
isisgreater
9-34
Thebars
barsindicate
indicatethe
thethree
threegroupings
groupingsof
ofpopulations
populationswith
withpossibly
possiblyequal
equal
The
means:22and
and3;3;2,2,3,3,and
and5;5;and
and1,1,4,4,and
and5.5.
means:
9-35
9-36
statisticalmodel
modelisisaaset
setof
ofequations
equationsand
andassumptions
assumptions
AAstatistical
thatcapture
capturethe
theessential
essentialcharacteristics
characteristicsof
ofaareal-world
real-world
that
situation
situation
Theone-factor
one-factorANOVA
ANOVAmodel:
model:
The
=++i+
i+ijij=
i+ijij
xxijij==i+
whereijijisisthe
theerror
errorassociated
associatedwith
withthe
thejth
jthmember
memberof
of
where
theith
ithpopulation.
population. The
Theerrors
errorsare
areassumed
assumedto
tobe
be
the
normallydistributed
distributedwith
withmean
mean00and
andvariance
variance22..
normally
9-37
factorisisaaset
setof
ofpopulations
populationsor
ortreatments
treatmentsof
ofaasingle
singlekind.
kind. For
For
AAfactor
example:
example:
Onefactor
factormodels
modelsbased
basedon
onsets
setsof
ofresorts,
resorts,types
typesof
ofairplanes,
airplanes,or
or
One
kindsof
ofsweaters
sweaters
kinds
Twofactor
factormodels
modelsbased
basedon
onfirm
firmand
andlocation
location
Two
Threefactor
factormodels
modelsbased
basedon
oncolor
colorand
andshape
shapeand
andsize
sizeof
ofan
anad.
ad.
Three
Fixed-Effectsand
andRandom
RandomEffects
Effects
Fixed-Effects
Afixed-effects
fixed-effectsmodel
modelisisone
oneininwhich
whichthe
thelevels
levelsof
ofthe
thefactor
factorunder
under
A
study(the
(thetreatments)
treatments)are
arefixed
fixedininadvance.
advance. Inference
Inferenceisisvalid
validonly
only
study
forthe
thelevels
levelsunder
understudy.
study.
for
Arandom-effects
random-effectsmodel
modelisisone
oneininwhich
whichthe
thelevels
levelsof
ofthe
thefactor
factor
A
understudy
studyare
arerandomly
randomlychosen
chosenfrom
froman
anentire
entirepopulation
populationof
oflevels
levels
under
(treatments). Inference
Inferenceisisvalid
validfor
forthe
theentire
entirepopulation
populationof
oflevels.
levels.
(treatments).
9-38
Experimental Design
completely-randomizeddesign
designisisone
onein
inwhich
whichthe
the
AAcompletely-randomized
elementsare
areassigned
assignedto
totreatments
treatmentscompletely
completelyat
atrandom.
random.
elements
Thatis,
is,any
anyelement
elementchosen
chosenfor
forthe
thestudy
studyhas
hasan
anequal
equal
That
chanceof
ofbeing
beingassigned
assignedto
toany
anytreatment.
treatment.
chance
Inaablocking
blockingdesign,
design,elements
elementsare
areassigned
assignedto
totreatments
treatments
In
afterfirst
firstbeing
beingcollected
collectedinto
intohomogeneous
homogeneousgroups.
groups.
after
Inaacompletely
completelyrandomized
randomizedblock
blockdesign,
design,all
allmembers
membersof
ofeach
each
In
block(homogeneous
(homogeneousgroup)
group)are
arerandomly
randomlyassigned
assignedtotothe
thetreatment
treatment
block
levels.
levels.
Inaarepeated
repeatedmeasures
measuresdesign,
design,each
eachmember
memberof
ofeach
eachblock
blockisis
In
assignedtotoall
alltreatment
treatmentlevels.
levels.
assigned
9-39
two-wayANOVA,
ANOVA,the
theeffects
effectsof
oftwo
twofactors
factorsorortreatments
treatmentscan
canbe
beinvestigated
investigated
InInaatwo-way
simultaneously. Two-way
Two-wayANOVA
ANOVAalso
alsopermits
permitsthe
theinvestigation
investigationofofthe
theeffects
effectsofof
simultaneously.
eitherfactor
factoralone
aloneand
andof
ofthe
thetwo
twofactors
factorstogether.
together.
either
The effect on the population mean that can be attributed to the levels of either factor alone
The
effect on the population mean that can be attributed to the levels of either factor alone
calledaamain
maineffect.
effect.
isiscalled
An interaction effect between two factors occurs if the total effect at some pair of levels of
An
interaction effect between two factors occurs if the total effect at some pair of levels of
the
twofactors
factorsorortreatments
treatmentsdiffers
differssignificantly
significantlyfrom
fromthe
thesimple
simpleaddition
additionofofthe
thetwo
twomain
main
the two
effects. Factors
Factorsthat
thatdo
donot
notinteract
interactare
arecalled
calledadditive.
additive.
effects.
Threequestions
questionsanswerable
answerableby
bytwo-way
two-wayANOVA:
ANOVA:
Three
Forexample,
example, we
wemight
mightinvestigate
investigatethe
theeffects
effectson
onvacationers
vacationersratings
ratingsof
ofresorts
resortsby
by
For
lookingatatfive
fivedifferent
differentresorts
resorts(factor
(factorA)
A)and
andfour
fourdifferent
differentresort
resortattributes
attributes(factor
(factor
looking
B). InInaddition
additiontotothe
thefive
fivemain
mainfactor
factorAAtreatment
treatmentlevels
levelsand
andthe
thefour
fourmain
mainfactor
factorBB
B).
treatmentlevels,
levels,there
thereare
are(5*4=20)
(5*4=20)interaction
interactiontreatment
treatmentlevels.3
levels.3
treatment
9-40
theeffect
effectof
oflevel
leveli(i=1,...,a)
i(i=1,...,a)of
offactor
factorA;
A;
i iisisthe
isthe
theeffect
effectof
oflevel
levelj(j=1,...,b)
j(j=1,...,b)of
offactor
factorB;
B;
j jis
jjjjisisthe
theinteraction
interactioneffect
effectof
oflevels
levelsiiand
andj;j;
isthe
theerror
errorassociated
associatedwith
withthe
thekth
kthdata
datapoint
pointfrom
from
jjk
jjkis
leveliiof
offactor
factorAAand
andlevel
leveljjof
offactor
factorB.
B.
level
isassumed
assumedto
tobe
bedistributed
distributednormally
normallywith
withmean
meanzero
zero
jjk
jjkis
andvariance
variance22for
forall
alli,i,j,j,and
andk.k.
and
9-41
Factor A: Resort
Friendship
Sports
Culture
Excitement
Guadeloupe
n11
n12
n13
n14
Martinique
n21
n22
n23
n24
Eleuthra
n31
n32
n33
n34
Rating
Friendship
R a ting
Excitement
Sports
Culture
Friendship
Paradise
Island
n41
n42
n43
n44
St. Lucia
n51
n52
n53
n54
Eleuthra/sports interaction:
Combined effect greater than
additive main effects
Attribute
Excitement
Sports
Culture
Eleuthra
St. Lucia
Paradise island
Martinique
Guadeloupe
Resort
Resort
St. Lucia
Paradise Island
Eleuthra
Guadeloupe
Martinique
9-42
Factor BB main
main effects
effects test:
test:
Factor
H0: j= 0 for all j=1,2,...,b
Test for
for (AB)
(AB) interactions:
interactions:
Test
9-43
Sums of Squares
In a two-way ANOVA:
In a two-way ANOVA:
=+i+
j j ++(
(ijkijk ++ijkijk
i+
xxijkijk=+
SST==SSTR
SSTR+SSE
+SSE
SST
SST==SSA
SSA++SSB
SSB+SS(AB)+SSE
+SS(AB)+SSE
SST
SST SSTR SSE
( x x )2 ( x x )2 ( x x )2
SSTR SSA SSB SS ( AB)
( x x )2 ( x x )2 ( x x x x )2
i
j
ij i
j
9-44
Sum of
Squares
Degrees
of Freedom
Mean Square
F Ratio
Factor A
SSA
a-1
MSA
SSA
a 1
MSA
F
MSE
Factor B
SSB
b-1
MSB
SSB
b 1
MSB
F
MSE
Interaction SS(AB)
(a-1)(b-1)
MS ( AB)
Error
SSE
ab(n-1)
Total
SST
abn-1
SS ( AB)
( a 1)(b 1)
SSE
MSE
ab( n 1)
MS ( AB)
F
MSE
9-45
Sum of
Squares
Degrees
of Freedom
Location
1824
912
8.94
Artist
2230
1115
10.93
804
201
1.97
Error
8262
81
102
Total
13120
89
Interaction
Mean Square
F Ratio
=0.01,FF(2,81)=4.88
=4.88
Both
Bothmain
maineffect
effectnull
nullhypotheses
hypothesesare
arerejected.
rejected.
=0.01,
(2,81)
=0.05,FF(2,81)=2.48
=2.48
Interaction
Interactioneffect
effectnull
nullhypotheses
hypothesesare
arenot
notrejected.
rejected.
=0.05,
(2,81)
9-46
Hypothesis Tests
F Distribution with 2 and 81 Degrees of Freedom
0.7
0.6
0.4
0.5
f(F)
f(F)
0.5
0.6
0.4
0.3
0.3
=0.01
0.2
=0.05
0.2
0.1
0.1
0.0
0
F0.01=4.88
0.0
0
F0.05=2.48
T q
MSE
bn
wherethe
thedegrees
degreesof
offreedom
freedomof
ofthe
theqqdistribution
distributionare
arenow
nowaaand
andab(n-1).
ab(n-1).
where
Notethat
thatMSE
MSEisisdivided
dividedby
bybn.
bn.
Note
9-47
9-48
9-49
Sum of
Squares
Degrees
of Freedom
Mean Square
SSA
a 1
Factor A
SSA
a-1
MSA
Factor B
SSB
b-1
SSB
MSB
b 1
Factor C
SSC
c-1
MSC
Interaction
(AB)
Interaction
(AC)
Interaction
(BC)
SS(AB)
(a-1)(b-1)
SS(AC)
(a-1)(c-1)
SS(BC)
(b-1)(c-1)
SS ( AB)
( a 1)(b 1)
SS ( AC)
MS ( AC)
(a 1)(c 1)
SS ( BC)
MS ( BC)
(b 1)(c 1)
Interaction
(ABC)
Error
SS(ABC)
(a-1)(b-1)(c-1)
SSE
abc(n-1)
Total
SST
abcn-1
SSC
c 1
MS ( AB)
SS ( ABC)
(a 1)(b 1)(c 1)
SSE
MSE
abc( n 1)
MS ( ABC)
F Ratio
MSA
F
MSE
F
MSB
MSE
MSC
MSE
MS ( AB )
F
MSE
F
MS ( AC )
F
MSE
MS ( BC)
F
MSE
MS( ABC)
F
MSE
9-50
9-51
Sum of
Squares
Degrees of
Freedom
Factor A
SSA
a-1
Factor B
SSB
b-1
Error
SS(AB)
(a 1)(b
1)
Total
SST
ab - 1
Mean Square
F Ratio
MSA SSA
a 1
F MSA
MS (AB)
MSB SSB
b 1
F MSB
MS (AB)
MS ( AB) SS ( AB)
(a 1)(b 1)
9-52
xxijij
mean zero
zero and
and variance
variance 22 for
for all
all ii and
and j.j.
mean
9-53
SSBL
SSTR
SSE
SST
Source of Variation
Blocks
Treatments
Error
Total
n-1
r-1
(n -1)(r - 1)
nr - 1
9-54
F Ratio
Sum of Squares
df
Mean Square F Ratio
2750
39
70.51
0.69
2640
2
1320
12.93
7960
78
102.05
13350 119
9-55