Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 23 October 2014
Received in revised form 4 January 2015
Accepted 13 February 2015
Keywords:
Solid waste pollution
Environmental concern
Solid waste management
Factor analysis
a b s t r a c t
Pollution concern has been established as an important predictor of economic value ascription related to
solid waste service improvement, such as recycling. However, the variable is not directly observable and
as such, its inuence is captured by existing applied economics papers via yes or no discrete-nominal
measure. Such measure assumes the variable has one dimension. This leads to loss of information, considering its multi-dimensional structure in theory. It thus implies that, if a respondent in an economic
valuation study indicates a support for an improvement signied by a yes response to valuation question, a uni-dimensional dummy variable cannot tell what dimension of concern the respondent subscribes
to. Despite this importance of understanding the variables dimension, it has rarely been explored. This
study investigates such dimensions and our ndings support the existence of both the theoretically
espoused dual and tripartite factor models. This implies a possible misspecication in existing solid
waste related studies that have most often captured the variables via uni-dimensional nominal-discrete
measure.
2015 Published by Elsevier B.V.
1. Introduction
Environmental concern is recently acknowledged as an important non-economic motivator of pro-environmental activities
which affect welfare estimates in economic valuation of environmental resources. This variable is broadly expressed in the context
of environmental concern (Ojea and Loureiro, 2007; Stern et al.,
1993; Stern and Dietz, 1994; Whittaker et al., 2005). Environmental concern is understood to have a few distinct dimensions and
all dimensions are believed to simultaneously exist in all humans,
motivating us to obtain utility from pro-environmental activities
even at some costs (Stern et al., 1993; Stern and Dietz, 1994;
Thompson and Barton, 1994).
Concern is a form of attitude or ethical consideration (Stern
et al., 1993; Stern and Dietz, 1994; Ojea and Loureiro, 2007).
However, economists have generally given little attention to the
consideration of all forms of attitudes, including concern, in
50
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Afroz et al. (2009), Afroz and Masud (2011), Jin et al. (2006),
and Othman (2003, 2007) assess the impact of solid waste pollution concern on willingness to pay (WTP) for waste management
service improvement. Meanwhile, Jamelske and Kipperberg (2006)
and Tadesse (2009) assess its impact on WTP for solid waste recycling service and willingness to separate solid waste. Similarly, Pek
and Jamal (2011) examine the impact of concern for solid waste
on WTP for the benets of solid waste alternative disposal options.
These studies have credibly examined the impact of environmental
concern specic to solid waste or particularly solid waste pollution
concern on economic value or WTP for solid waste improvement
or willingness to separate solid waste.
However, all aforementioned studies capture solid waste pollution concern as a YesNo discrete variable, allowing a great
deal of information to be lost. This is because such conceptualization ignores the distinct dimensions of the variable. In theory,
at least, two to three dimensions of such concern exist synchronously in every individual but not all such dimensions may
be activated to inuence solid waste management related proenvironmentalism such as paying for its improvement (Stern and
Dietz, 1994; Thompson and Barton, 1994). Thus, pollution concern
data collected based on YesNo discrete measure as presently
adopted in existing studies provide no information on what dimension of its structure is activated in support of value ascription. At
best, it only gives insight into the relationship between economic
value and concern. Meanwhile, information on such is important to inform policy on what dimension to focus when designing
awareness campaign programs to trigger support for solid waste
management improvement or related policies.
To address the above issue, this study assesses the dimensions
of solid waste pollution concern that could affect its related value
ascription by individuals. As far as our knowledge goes, this study is
the rst to examine such dimensions in the context of solid waste
pollution concern. The rest of the paper is organized as follows.
Section 2 discusses the dimensions of concern based on relevant
frameworks, followed by discussion on the method of analysis in
Section 3. Section 4 presents the studys results and discussion, and
Section 5 concludes.
believed there are at least three value dimensions that underlie environmental concern, attitude or environmentalism. They
noted that egoistic value which emphasizes environmental concern evoked by the motivation to protect the environment for the
benet of ones self is one of the value orientations that underlie concern. Secondly, drawing from Schwartzs theory of altruism
(Schwartz, 1968a, 1968b, 1970, 1977) they established that there
is a social or altruistic motivation behind environmental concern.
This entails holding a concern in favor of the environment solely to
enhance the welfare of other human beings.
The third orientation they perceived is the value orientation
toward non-human species or the biosphere itself as propagated
by ecologists such as the land ethic by Leopold (1948) as well
as environmentalists (Stern and Dietz, 1994). Biospheric concern
as instrument items appeared early in the works of Dunlap and
Liere (1977a, 1977b) and Heberlein (1972, 1977). This is conceived
within the framework of Stern et al. (1993) as a form of altruism that
transcends human welfare to the welfare of non-human species.
This implies some conceptual correlations between biospherism
and altruism.
Stern et al. (1993) noted that the three aforementioned value
orientations of environmental concern are not held in exclusivity.
Rather, they may be compatible and peoples environmental concern reects some mixture of the three orientations. However, the
authors obtained a statistically insignicant correlation between
egoistic and biospheric values.
Alternatively, Thompson and Barton (1994) introduced a similar
but slightly different conceptualization of environmental concern
structure based on anthropocentricecocentric distinctions. This
theoretical framework posits that there are at least two value orientations underlying concern for environmental issues. As with
Stern et al. (1993), they concluded that individuals holding either or
both orientations engage in pro-environmental activities, but the
purpose of their engagement differs. Ecocentric individuals value
the nature only for its sake and protect it based on its intrinsic
value. Meanwhile, anthropocentrics engage in pro-environmental
activities on account of its value to enhance the quality of life
for humans (Thompson and Barton, 1994). Anthropocentrism is
human-centered while ecocentrism emphasizes the wellbeing of
nature even at the expense of material loss.
Hence, they pointed that the motives are distinguishable which
would suggest low conceptual correlation among the dimensions.
In a further elaboration, Thompson and Barton (1994) compared
their framework of environmental concern to the earlier model
by Stern et al. (1993). They pointed that both egoistic and socialaltruistic values in the earlier theoretical framework are equivalent
to their concept of anthropocentric value because they focus on
outcomes for humans. Whereas, they noted that, biospheric value
in the earlier framework is equivalent to their ecocentric motive.
Thus under Thompson and Bartons (1994) framework, we expect
a convergence between egoistic and social-altruistic values to form
anthropocentric value while a low correlation is expected between
the latter and the ecocentric value. It is therefore not surprising that
their study reported a non-signicant correlation between these
two dimensions of anthropocentric and biospheric (ecocentric) values.
A few studies have attempted conrming the dimensions of general environmental concern by using a reduced version of existing
scales. These include Amrigo et al. (2005, 2007) and Straughan
and Roberts (1999). The present study aims at determining the
dimensions of solid waste pollution concern. We rely on the two
prominent frameworks of concern presented in Stern et al. (1993),
Thompson and Barton (1994) and Amrigo et al. (2007). These
two theories are presented in Fig. 1. It is instructive to note that
the items used to demonstrate the concern dimensions in these
previous studies are generic, comprising myriad of environmental
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Tripartite theory of
Value/Attitude/ Concern
51
Dualistic theory of
Value/Attitude/ Concern
Egoistic Concern
Anthropocentric
Concern
Environmental
(Pollution)
Concern
Social-Altruistic
Concern
Ecocentric
Concern
Biospheric
Concern
3.1. Participants
The respondents for this study were household representatives
in urbanslum areas of Lagos state in Nigeria. Slums are settlements characterized by inadequate housing and basic services
(UN-HABITAT, 2007). These characteristics dene the suburbs of
Lagos, which arises as a result of the citys high rate of urbanization
which causes urban sprawl (Aluko, 2010; Felix, 2007). Usually rural
areas are considered as areas outside urban metropolitan region.
However, in places characterized by high rate of urbanization leading to the scenario of urban sprawl, the city-center expands until
it engulfs its suburbs (Felix, 2007). This is the case of Lagos, which
Sectors
Urban areas
Count
% within sector
% within disposal type
% of total
Rural areas
Count
% within sector
% within disposal type
% of total
Total
Improper disposal
Proper disposal
53
15.2%
37.9%
11.8%
296
84.8%
95.5%
65.8%
349
100%
77.6%
77.6%
87
86.1%
62.1%
19.3%
14
13.9%
4.5%
3.1%
101
100%
22.4%
22.4%
52
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Table 2
Test items for environmental concern in the questionnaire.
Egoistic concern items
1.
I am happy to be in an environment
clean of solid waste pollution (2)
5.
2.
6.
3.
7.
4.
8.
10.
11.
12.
Survey questionnaire.
(i = 1, 2, . . ., p)
(1)
Eq. (1) denotes a factor analysis for the vectors of X observed variables and F underlying common factors such that:
X = (x1 , x2 , . . ., xp ), F = (f1 , f2 , . . ., fq ) and is a constant (called
factor loadings/weight) denoting the correlation between individual components of X and F. In a more concise form capturing the
vectors of X, F, and E, Eq. (1) could be re-specied as:
X = F + E
where X is a matrix of the measured items/variables, F is a matrix
of underlying structure (common factors), is a matrix of factor
loadings, E is a matrix of error variance (unique factors).
Based on this specication, factor analysis permits the location
of each of the sampled individuals within the factor space (F-space)
based on their observed values (i.e., responses to questionnaire
items X). Thus, in the specic application of the present study, X
denotes all the 12 individual indicators of solid waste pollution
dimensions (xi ) while F represents the three or two dimensions
(fi ) respectively hypothesized under the tripartite and dual factor
models.
3.5. Analysis procedure
The data obtained was subjected to exploratory factor analysis (EFA) to respectively determine the dimensional structure for
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
53
Table 3
Respondents summary statistics.
Socioeconomic variables
Gender
Male
Female
Marital status
Single
Married
Others
Age
Years of educational attainment
Never been to school
Primary school
Secondary school
N.C.E/diploma
University/HND
Occupation
Civil servant
Private sector
Pension
Self-employed
Housewife
Unemployed
Number of households in apartments
Monthly family income ( )
10,000 50,000
51,000
101,000
>
100,000
200,000
200,000
Frequency
Percentage
162
138
54
46
65
229
6
21.7
76.3
2
75
23
82
82
38
25
7.7
27.3
27.3
12.7
27
65
9
156
17
26
9
21.7
3
52
5.7
8.7
158
52.7
110
36.7
31
10.3
0.3
Mean
Std. deviation
0.54
0.49
35.43
9.87
8.99
6.23
12.96
58,333.3
4.95
39,132.6
http://www.unicef.org/infobycountry/nigeria statistics.html.
This is based on the exchange rate of 1 USD =
160.
This estimate is provided by the National Bureau of statistics, Nigeria. It is available at: http://nigeria.prognoz.com/en/DataAnalysis.
2
3
54
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Table 4
Summary statistics, reliability and univariate normality for attitudinal concern items.
Attitudinal concern dimensions (n)
Mean
Egoistic concern
EGO1
EGO2
EGO3
EGO4
4.44
Biospheric concern
BIO1
BIO2
BIO3
BIO4 R
2.64
Social-altruistic concern
SOCTR1
SOCTR2
SOCTR3 R
SOCTR4
Observations
4.17
Std. deviation
Skewness
Kurtosis
0.814
0.623
0.687
0.659
0.666
0.891
0.705
0.713
0.743
0.227
0.651
0.548
0.536
0.894
0.851
0.822
0.826
0.301
0.579
0.539
0.363
0.751
0.146
0.072
0.182
0.666
0.704
0.743
0.698
0.624
0.424
0.12
0.214
0.653
0.919
0.505
0.939
0.77
0.75
0.75
0.78
0.776
0.714
0.732
0.695
0.747
0.732
0.702
0.613
0.654
0.712
234
Inverse Normal
Inverse Normal
represent distinct dimensions. This is necessary to justify aggregation under each sub-dimension for further analysis (Barr, 2007).
As suggested by Santos (1999), reliability analysis aims at testing
the internal consistency of each dimension and it precedes factor
analysis.
4
Inverse Normal
Inverse Normal
55
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
2
3
Inverse Normal
2
3
Inverse Normal
2
3
Inverse Normal
4
Inverse Normal
Reliability measures the extent to which multiple-choice itemsscores obtained from questionnaire responses for each underlying
variable are free from random measurement error (Kline, 2005).
There are several tests of reliability, including; the testretest,
split-half reliability, Guttman reliability, Parallel reliability, strictly
3
Inverse Normal
4
Inverse Normal
56
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
( = 0.73) were all reliable. This implies that the items for each
of the latent sub-dimensions of concern are internally consistent.
This provides a good justication for analyzing the data from the
questionnaire (Barr, 2007).
4
Inverse Normal
Cronbach alpha coefcients are reported. The standardized coefcients are obtained from the average correlation of standardized
items-scores within a dimension. Meanwhile, unstandardized
coefcients are computed based on the average covariance itemsscores within a dimension. Cronbachs alpha could be interpreted
similar to correlation coefcient with its value ranging from 1 to 0.
When the scales for measuring a particular variable, such as the
concern for solid waste pollution, contain sub-scales (dimensions)
that may not necessarily be combined, it is more appropriate to
calculate reliability for each dimension separately (Pallant, 2010).
Reliability coefcient is sensitive to both sample size and the number of items that form each scale. As such, it is not uncommon
to nd coefcients as low as 0.5 (Pallant, 2010). While no gold
standard exist as to how high the reliability coefcient should be,
there is general consensus. Reliability coefcients 0.90 is considered excellent, a value 0.80 is deemed very good, while values
0.70 are considered ideal (Kline, 2005; Pallant, 2010).
Accordingly, the reliability of sub-dimensions of concern was
assessed based on Cronbachs alpha. All individual items reliability ranges from 0.61 to 0.78, as shown in Table 4. Their coefcients
were found to be less than the overall computed reliability value
for each dimension. This suggests the appropriateness of all items
(Coakes et al., 2009; Pallant, 2010). On the other hand, the reliability
for each aggregate dimension ranges from 0.73 to 0.81. Egoistic concern ( = 0.81), biospheric concern ( = 0.77), and social-altruistic
4
Inverse Normal
Exploratory factor analysis (EFA) and conrmatory factor analysis (CFA) are the two major types of analyses based on the statistical
common factor model (Brown, 2006, 2012; Cudeck and MacCallum,
2012; Thompson, 2004). However, EFA is used in this study since
the objective is only to explore the underlying dimensions of
concern. The indispensable issue of concern under EFA is the factorability of items. This could be disaggregated into sample size
requirement, sufcient inter-correlation both for individual items
and the entire items as a whole and the avoidance of outliers
(Pallant, 2010; Coakes et al., 2009). The rst factorability prerequisite of sample size requirement is met since the ratio of indicator
to free parameter exceeds 1:10 as evidenced by the sample size
of 234 for a maximum of three factors (Westland, 2010; Nunnally,
1978).
In the case of the second factorability requirement aforementioned, Tabachnick and Fidell (2006) suggested the inter-items
correlations to be examined for the dominance of associations
greater than 0.3. The inter-items correlations result shown in Table
1 in Appendix, evidences the fulllment of this requirement. Furthermore, the factorability requirement of sample adequacy for
individual items was examined. This was conrmed via the antiimage correlation matrix which is used to determine the sampling
adequacy of each item (variable). As shown in Table 5, the values
obtained for the items in this study range from 0.69 to 0.88. These
values, which are shown along the diagonal of Table 5, are greater
than the least acceptable threshold of 0.5 (Coakes et al., 2009; Hair
et al., 2010). This implies that all items are sufciently sampled to
permit factor analysis.
As for the last factorability requirement of overall items sample
adequacy, the KMO as well as Bertletts test of sphericity, reported
in Table 5 were found within acceptable thresholds. This justies the factorability of items. Since KMO value is greater than the
threshold of 0.6 and Bertletts test of sphericity is less than 0.05, the
factorability of all items is this study is assumed appropriate.
The next important item entails researchers employment of factor extraction method. There exist a long list of methods including
principal components, principal factors, image factoring, maximum
likelihood factoring, alpha factoring, unweighted least square and
generalized least square (Pallant, 2010). However, the most common approach (principal components) is applied in this study.
To determine the exact number of factors to be extracted, three
approaches were employed. These include the Kaiser criterion, the
scree plot and parallel analysis. While the rst two are directly
included in SPSS output, the third technique proposed by Horn
(1965), is not generated in the output. Thus, it was computed using
Monte Carlo PCA software for parallel analysis (Watkins, 2000).
The computation of result based on this last method is considered very important as both the Kaiser criterion and scree plot
potentially overestimate the true number of components (Pallant,
2010). To obtain an interpretable factor pattern, direct Oblimin
(oblique) rotation was used to examine which items converge
together. This choice of rotation method is based on the theoretical expectation that concern dimensions have correlated structural
dimensions.
To determine the appropriate number of factors to extract, it
is necessary to refer to the magnitude of eigenvalue greater than
one in Table 6 (Kaiser criterion). Another criterion entails observing the number of factors above the elbow of Cattells (1966) scree
plot shown in Fig. 1 of Appendix. Although determining the point
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
57
Table 5
Overall and individual items measure of sample adequacy (MSA).
Individual items measure of sample adequacy
EGO1
EGO2
EGO3
EGO4
BIO1
BIO2
BIO3
BIO4 R
SOCTR1
SOCTR2
SOCTR3 R
SOCTR4
0.86a
0.30
0.13
0.18
0.03
0.06
0.03
0.13
0.00
0.15
0.06
0.08
0.85a
0.36
0.01
0.08
0.05
0.03
0.07
0.09
0.05
0.05
0.18
0.82a
0.31
0.04
0.04
0.01
0.09
0.02
0.01
0.17
0.10
0.87a
0.07
0.04
0.05
0.04
0.21
0.11
0.02
0.05
0.70a
0.00
0.29
0.45
0.13
0.02
0.09
0.11
0.70a
0.53
0.16
0.08
0.02
0.02
0.04
0.69a
0.03
0.00
0.09
0.01
0.06
0.71a
0.010
0.01
0.02
0.03
0.88a
0.26
0.05
0.04
0.86a
0.11
0.31
0.87a
0.25
0.83a
952.9
66
0.0000
Table 6
Total variance explained and components extraction based on eigenvalues.
Component
1
2
3
4
5
6
7
8
9
10
11
12
Initial eigenvalues
Total
% of variance
Cumulative %
Total
% of variance
Cumulative %
3.892779
2.501591
0.935763
0.842889
0.736322
0.628709
0.510187
0.490404
0.420386
0.398045
0.335388
0.307537
32.43983
20.84659
7.798029
7.024075
6.136016
5.239243
4.251556
4.086702
3.503215
3.317038
2.794901
2.562809
32.43983
53.28642
61.08445
68.10852
74.24454
79.48378
83.73534
87.82204
91.32525
94.64229
97.43719
100
3.892779
2.501591
0.935763
32.43983
20.84659
7.798029
32.43983
53.28642
61.08445
where the elbow begins on the scree plot can be quite subjective,
the eigenvalue suggests only two factors extraction to explain solid
waste pollution concern. Together, the two factors explain 53.3% of
variation in households solid waste pollution concern. It could be
reasonable to consider the third factor which has an eigenvalue
close to 1 (i.e., 0.9). This increases the total variance explained to
about 61%. Yet, the parallel analysis result in Table 7 would not
suggest such. This is because factors beyond the rst-two are all
rejected. Hence, using a strictly exploratory EFA, we could conclude
that the result obtained supports the existence of a dual-concern
Table 7
Parallel analysis result.
Component number
Decision rule
1
2
3
4
5
6
7
8
9
10
11
12
3.892779
2.501591
0.935763
0.842889
0.736322
0.628709
0.510187
0.490404
0.420386
0.398045
0.335388
0.307537
1.3849
1.2809
1.2036
1.1345
1.0727
1.0128
0.9592
0.9043
0.8507
0.7941
0.7368
0.6656
Accept
Accept
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
Reject
58
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Table 8
Two-factor structure for solid waste pollution concern solution.
Questionnaire items
Pattern matrixa
Structure matrixa
Anthropocentric concern
EGO2
SOCTR2
EGO1
EGO4
EGO3
SOCTR4
SOCTR1
SOCTR3 R
BIO3
BIO1
BIO2
BIO4 R
Biospheric concern
Communalitiesa
Anthropocentric concern
0.771
0.737
0.728
0.722
0.714
0.675
0.627
0.57
Biospheric concern
0.809
0.786
0.761
0.714
0.599
0.567
0.534
0.547
0.515
0.462
0.451
0.337
0.657
0.626
0.585
0.514
0.77
0.739
0.727
0.721
0.713
0.674
0.629
0.571
0.808
0.787
0.76
0.715
Correlation
Anthrop
Biosph
Anthrop
Biosph
0.01
1
Factors are extracted using principal component analysis (PCA) based on oblique rotation with Kaiser normalization.
Table 9
Three-Factor Structure for Solid Waste Pollution Concern solution.
Questionnaire items
Pattern matrixa
Ego
EGO3
EGO2
EGO4
EGO1
BIO1
BIO3
BIO4 R
BIO2
SOCTR4
SOCTR2
SOCTR3 R
SOCTR1
Structure matrixa
Bio
Altr
0.902
0.736
0.719
0.648
Ego
Communalitiesa
Bio
Altr
0.812
0.782
0.648
0.573
0.563
0.654
0.733
0.606
0.654
0.612
0.664
0.610
0.457
0.655
0.693
0.430
0.849
0.799
0.767
0.732
0.807
0.780
0.765
0.716
0.798
0.796
0.742
0.740
0.808
0.678
0.600
0.411
Correlation
Ego
Bio
Altr
Ego
Bio
Altr
0.031
1
0.445
0.068
1
Factors are extracted using principal component analysis (PCA) based on oblique rotation with Kaiser normalization.
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
59
60
S.U. Adam et al. / Resources, Conservation and Recycling 104 (2015) 4960
Schwartz SH. Words, deeds and the perception of consequences and responsibility
in action situations. J Personal Soc Psychol 1968b;10(3):232.
Schwartz SH. Moral decision making and behavior. Altruism Help Behav
1970:12741.
Schwartz SH. Normative inuences on altruism. Adv Exp Soc Psychol
1977;10:22179.
Stern PC, Dietz T. The value basis of environmental concern. J Soc Issues
1994;50(3):6584.
Stern PC, Dietz T, Guagnano GA. The new ecological paradigm in socialpsychological
context. Environ Behav 1995;27(6):72343.
Stern PC, Dietz T, Kalof L. Value orientations, gender, and environmental concern.
Environ Behav 1993;25(5):32248.
Stern PC, Dietz T, Guagnano GA, Kalof L. A Value-Belief-Norm Theory of Support
for Social Movements: The Case of Environmentalism. Human Ecology Review
1999;6(2):8197.
Straughan RD, Roberts JA. Environmental segmentation alternatives: a look
at green consumer behavior in the new millennium. J Consum Market
1999;16(6):55875.
Tabachnick BG, Fidell LS. Using multivariate statistics. New York: Harper and Row;
2006.
Tadesse T. Environmental concern and its implication to household waste separation and disposal: evidence from Mekelle, Ethiopia. Resour Conserv Recycl
2009;53(4):18391.