ENVIRONMENTAL CONCERN AND PRO-ENVIRONMENTAL BEHAVIORS:
ATTITUDES AND BEHAVIORS IN URBAN CHINESE RESIDENTS
By
CHUNTIAN LU

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of
DOCTOR OF PHILOSOPHY
Sociology
2012

ABSTACT
ENVIRONMENTAL CONCERN AND PRO-ENVIRONMENTAL BEHAVIORS:
KNOWLEDGE, ATTITUDES AND BEHAVIORS IN RUBAN CHINESE RESIDENTS
By
CHUNTIAN LU
Since 1978, China‟s rapid economic growth has also generated numerous environmental
problems, which cause substantial economic losses, social conflict, and rising health costs. This
requires us to examine the factors affecting attitudes and behaviors toward the environment
among the Chinese. This study used data from 2003 General Social Survey in China to answer
the following questions: Is the measurement of environmental concern in China different from in
western societies? Will the selection of particular measure of environmental concern affect its
relation to social-demographic variables? What are the effects of individual and community level
variables on two types of environmental behaviors?
The results showed that environmental concern in China includes four dimensions:
economical trade off, pro-environmental behaviors, perceived locally environmental problems
and NEP. However, the results are somewhat different from those in western societies. The
analysis suggests that more attention should be paid to the selection of particular measure of
environmental concern and demonstrates that it really makes a difference how environmental
concern is measured. Our findings indicate that except for gender and income, the effects of other
individual variables follow the same pattern as in the West. As for municipal level variables, our
analysis demonstrates that the affluence hypothesis, which states that the level of national income
per capita influences the level of environmental concern, is partially supported while the objective
hypothesis, which argues that the rising of environmental concern is consequence of deteriorating
objective environmental problems, is not fully supported here.

This research provides some guidelines for future study of environmental concern and fills an
important gap in the environmental attitudes and behaviors research in China.

ACKNOWLEDGMENT
I owe my gratitude to all those people who have made this dissertation possible and because
of whom my graduate experience has been one that I will cherish forever.
My deepest gratitude is to my major advisor, Dr. Stan Kaplowitz. I would like to sincerely
thank Dr. Kaplowitz for his intellectual guidance and strong support whenever and for whatever
I need. His extensive experience in quantitative methodology and insightful understanding on
attitudes and behavior exhibit me an exciting world for exploring. I am always impressed by his
enthusiasm on research, his hard working, and his visions on new research fields.
Thanks from my heart also go to Dr. Zhenmei Zhang. I am grateful to her for providing me
opportunities to touch on various projects. Her instructions help me enrich but focus my ideas at
different stages. I also appreciate her support all the time, and her care about my living at school.
I would like to extend my special thanks to my committee members, Dr. Sandra Marquart-Pyatt
and Dr. Aaron McCright. Their insightful critics, constructive comments, guidance, and careful
reviews on my dissertation help me overcome difficulties and finalize my research. I am also
grateful for their kind understanding, patience and encouragement when I was slow of going
forward.
I am also grateful to faculty and staff at the Department of Sociology. I learned a lot from
classes, seminars and discussions; and helps from staff make it so comfortable studying at the
Department. I would also like to express gratitude to my graduate colleagues and friends at
MSU. Their care and friendships make my staying and working at MSU an enjoyable and
awarding experience.
Finally, and the most importantly, I would like to thank my parents Yongkang Lu and
Caiying Huang, my sister Chunyu Lu, and my brothers Chunhui Lu and Chunfeng Lu and their

iv

families. They are there for me all the time and home is always the place I most dream to go. My
family accompanied me going through all ups and downs, and I would not ask more than they
always give me: love, care, encouragement, strength and happiness.

v

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................................... viii
LIST OF FIGURES ....................................................................................................................... ix
Chapter 1 Introduction .....................................................................................................................1
1.1 Background ........................................................................................................................... 1
1.2 The Purpose of This Study.................................................................................................... 4
1.3 Research Questions ............................................................................................................... 6
1.4 Dissertation Outline .............................................................................................................. 7
BIBLIOGRAPHY ..................................................................................................................... 10
Chapter 2 Environmental Concern in China ..................................................................................13
2.1 Background ......................................................................................................................... 13
2.2 A Brief Introduction of Environmental Concern ................................................................ 15
2.3 Literature on the Dimensions of Environmental Concern .................................................. 17
2.4 Data and Method ................................................................................................................. 22
2.5 Measurement Model of Environmental Concern ................................................................ 24
2.6 Result and Discussion ......................................................................................................... 31
2.7 Conclusion .......................................................................................................................... 37
BIBLIOGRAPHY ..................................................................................................................... 40
Chapter 3 Social Demographic Characteristics of
Environmental Concern .................................................................................................................44
3.1 Introduction ......................................................................................................................... 44
3.2 Literature Review................................................................................................................ 46
3.3 Data and Method ................................................................................................................. 52
3.4 Results ................................................................................................................................. 58
3.5 Discussion ........................................................................................................................... 61
3.6 Conclusion and Summary ................................................................................................... 64
BIBLIOGRAPHY ..................................................................................................................... 66
Chapter 4 Multi-level Analysis of Urban Chinese Residents‟ Environmental Attitude and
Behaviors .......................................................................................................................................71
4.1 Introduction ......................................................................................................................... 71
4.2 Literature Review................................................................................................................ 73
4.2.1 Individual-Level Variables: Socio-demographic Characteristics ................................ 74
4.2.2 Individual Level Variables: Socio-psychological Characteristics ............................... 77
4.2.3 Contextual -Level Variables ........................................................................................ 78
4.3 Data Sources ....................................................................................................................... 81
4.3.1 Variables Measurement: Dependent Variables ............................................................ 82
4.3.2 Independent variables .................................................................................................. 84
4.5 Analysis Strategy ................................................................................................................ 88
vi

4.6 Results ................................................................................................................................. 91
4.7 Discussions and Conclusion ............................................................................................... 99
BIBLIOGRAPHY ................................................................................................................... 105
Chapter 5 Conclusion and Discussion .........................................................................................110
5.1 Overview ........................................................................................................................... 110
5.2 Summary of Research Findings ........................................................................................ 110
5.3 Significance and Limitations of the Study ........................................................................ 113
BIBLIOGRAPHY ................................................................................................................... 116

vii

LIST OF TABLES
Table 2-1 Discription of Indicators ........................................................................................ …..30
Table 2-2 Unstandardized (Standardized) Factor Loading from CFA ........................................ 32
Table 2-3 Correlation among these Four Factors ........................................................................ 34
Table 3-1 Three Items of Pro-environmental Behaviors ............................................................. 55
Table 3-2 Seven Items of the Perceived Local Environmental Problems ................................... 56
Table 3-3 Six Items of New Environmental Paradigm Scale Items and Percentage Frequencies in
China and U.S……………………………………………………………………………...56
Table 3-4 Social Demographic and Economic Variables ............................................................ 58
Table 3-5 Standardized Regression Coefficients on Each of Four Factors……………………..59
Table 4-1 Six New Environmental Paradigm Scale Items and Percentage Frequencies ............. 82
Table 4-2 Ten Items of Environmental Behaviors and Percentage Frequencies ......................... 84
Table 4-3 Seven Items of the Perceived Local Environmental Problems and Percentage
Frequencies .......................................................................................................................... 85
Table 4-4 Descriptions of City Level Variables ......................................................................... 87
Table 4-5 City Level Variables Correlation Matrix ..................................................................... 87
Table 4-6 Multi-level Analysis of Environmental Attitudes ....................................................... 92
Table 4-7 Multi-level Analysis of Private Environmental Behaviors ......................................... 95
Table 4-8 Multi-level Analysis of Public Environmental Behaviors........................................... 98

viii

LIST OF FIGURES

Figure 2-1 Measurement Model of Environmental Concern ....................................................... 28
Figure 3-1 Socio-demographic characteristics of Environmental Concern ................................. 53

ix

Chapter 1 Introduction
1.1 Background
Since the start of economic reform in 1978, China‟s rapid economic growth has become a
major driving force in the global economy with an average annual GDP growth rate of 8%,
which has lifted nearly 400 million people out of poverty between 1979 and 2005. However,
China‟s rapid economic growth has also generated numerous major environmental problems.
These problems not only threaten the health and prospects of current and future generations, but
also undermine the sustainability of long-term growth. For example, according to a report in
th

People Daily on Aug 11 , 2009, more than 300 children suffered from lead poisoning that was
likely caused by pollution from a smelter in Shanxi Province. This incident was only one of a
recent string of pollution disasters in China that have prompted social discontents and violent
protests in some areas. News reports are showing that such incidents have become common in
other provinces of China.
Current severe environmental degradation in China cannot wholly be attributed to economic
policies and development strategies in the post-reform period. It is also related to economic
policies and development philosophy of Chinese Communist Party in the era of Mao Zedong,
who controlled absolute ruling power in China until he died in 1976. Under Mao, the traditional
Chinese ideal of “harmony between heaven and humans” was abrogated and in favor of Mao‟s
insistence that “man must conquer nature.” Then, nature was the enemy to be conquered, but it
was not the only one. Anyone who disagreed with Mao was also viewed as a public enemy too.
According to Shapiro‟s research (Shapiro,2001), a series of utopian mass development campaign
such as Great Leap Forward aiming to achieve unrealistic steel production, led to ecosystem
collapse in many of China‟s rural areas and create a famine that claimed 30 million lives. During
1

Mao‟s period, attempts to conquer nature reached the peak and environmental degradation in
effect became a state policy (Shapiro, 2001).
The past command economy established in the Mao time has been abandoned and
transformed to a market orientated economy during the reform period.
However, China is still regarded as an authoritarian state. This authoritarian state has made a
deliberate effort to pursue economic growth while maintaining the rule of the Chinese
communist party. This effort is evidenced by its decisions on issues such as large construction
projects.

One obvious example is the construction of the Three Gorges Dam. Although

concerned scientists and scholars gave the warning that the project might be an ecological and
social disaster, the Chinese government regarded this dam as a sign of the advantages of
socialism. Political repression and forced relocations in the Three Gorges Dam project have
demonstrated that the Chinese government has not learned the lessons from the past causalities
of the wars against nature. The Three Gorges Dam is just one of a numerous big projects during
the period of the market transition of China. Today, more and more Chinese people are
involuntarily involved in the side effects of these projects due to combination of command
economy and authoritarian state.
China‟s environmental damage is not restricted to its borders, as its environmental problems
affect other countries. China is now the largest emitter of global greenhouse gasses. It was
reported that its dust and aerial pollutants are transported eastwards to neighboring countries and
even to North America. China also surpassed the United States to become the largest source of
carbon dioxide, which is a main factor causing global warming. China accounts for 33% of
global fish and seafood consumption and 15% of the world fish catch (Liu & Diamond, 2005).

2

As a „world factory‟, China not only exports products around the world but also consumes and
imports raw materials and natural resources. By doing so, it leaves pollutants in its wake.
At the same time, other countries also impact China‟s environment through globalization and
pollution transferring. In 1980, China produced 1.0% in world trade exports, while in 2003 China
produces 5.8% in world trade exports (Rumbaugh and Blancher, 2004). Such fast increases also
let China overuse its internal natural resources and raw materials from other countries. China
holds 37% of total exports to the advanced countries market, and these countries include the
United States, Japan and European Union (Rumbaugh and Blancher, 2004)
Increasing international exports means advanced countries also contribute to environmental
problems in China. In 2006, China attracted the most foreign direct investments among
developing countries, ranking the third worldwide. For the foreign investors from advanced
countries or areas, maximizing their profit is their interest, Therefore, they tend to transfer their
investments to countries with weak environmental protections to locate their polluting facilities
and will use their influences to make those protections weak (Shandra, John M, Bruce London,
Owen P Whooley, and John B Williamson, 2004). Although China‟s per capita environmental
impact is still far below that of developed countries, there is no doubt that the increase in total
human impacts on the world‟s environments will be enormous if China‟s per capita impact
catches up with middle-income level countries, let alone the US. Globalization may be partly to
blame for this situation, but it is hardly the only explanation.
There is a growing literature on the nature and scale of environmental destruction in China
(World Bank & SEPA, 2007). The environmental problems includes air pollution, concerns over
biodiversity, water pollution, and desertification, salinization, soil erosion and the accumulation
of trash. All of these problems cause substantial economic losses, social conflict, and rising

3

health costs in China. According to a recent report published in 2008, environmental problems
had become regarded as a top ten social problem in China (Chinese Environmental and Cultural
Advance Association, 2008). It is estimated that in the next ten years, the success of sustainable
development in China depends on whether the country can curtain this trend of environmental
degradation.
The spread of global, national and local environmental problems in the 1980s has led the
Chinese government and its citizens to recognize the importance of nature and the need for its
protection. For example, when the United Nations Conference on Environment and Development
(UNCED) was held in Rio de Janeiro in1992, the Chinese government launched a call for a
massive environmental education campaign. The national postal service of China even published
a series of stamps to mark the importance of this meeting. Since that meeting, many new terms
linked to the environment, such as acid rain, bio-diversity and sustainable development have
gradually come to be accepted by most Chinese. Therefore, protection of the natural environment
has become an emergent and important task within the national agenda as well as a focal point
for academic research.
1.2 The Purpose of This Study
Human behavior is the product of individual and collective values. With regards to
environmental degradation in China, some scholars try to seek answers from the traditions and
values which underpinned the action logics behind the Chinese behaviors. Harris (2004), for
example, argued that some traditional values which may offer benign guidelines for
environmental protection are impotent when placed against new consumption habits that mimic
those in the West. Peterson (2001: 96) has argued that “attitudes toward the environment in
China are not Western in origin, but stem from values and belief rooted in such religious

4

institutions as Buddhism, Confucianism and Taoism”. However, Li (1998) believes that Confucianism, despite any non-anthropocentric orientation, can help to mitigate the recent trend of
environmental destruction. Scholars and policymakers have offered several solutions to
environmental pollutions in China. Some believe that technical and economic development will
eventually solve environmental problems as proponent of environmental Kuznets curve propose.
Others suggest that humans should pay attention to current knowledge, attitudes and behaviors
related to the environment. That is, human attitudes and behaviors must be changed to be
compatible with the environment. Instead of seeking an explanation buried in past traditions and
value, I focus on the state of environmental concern and behaviors rooted in contemporary
Chinese daily life.
Current environmental degradation requires us to take a close look at the factors affecting
knowledge of, and attitudes and behaviors toward the environment among the Chinese. Harris
(2006) found that Chinese people tended to hold anthropocentric viewpoints of and instrumental
values toward, , the natural world. That is, nature is meant to serve humans. With this in mind, it
is not surprising that economic growth takes priority over environmental protection. Due to its
large population and the scale of its current economic development, Harris (2006:5) concluded
“one would be hard pressed to find a more explicit and profound example of how human
behavior can adversely affect the ecological environment than the ongoing experience of China.”
Measuring the level of environmental concern, attitudes, and behaviors among the Chinese
will be extremely important to develop a new approach to environmental protection. Without
having environmental concern, it is difficult for the public to understand the need for sustainable
environmental

behaviors.

Therefore,

developing

5

a

conceptual

framework

explaining

environmental concern is also important to understand the underlying forces related to
environmental behaviors in China.
1.3 Research Questions
There has been much research done on environmental concern and pro-environmental
behaviors in developed countries. However, very little effort has been expended on systematic
investigations of environmental attitudes and different types of environmental behaviors in
China. Several scholars have pointed out that determinants of environmental views and
behaviors vary between countries and cultures (Hunter 2000, Marguart-Pyatt 2007). Considering
the huge differences between China and most western countries, not only in their political
structure but also in their cultures, it is time to study environmental concern and its causes in the
social context of China and reveal its social differences of individual environmental concerns
and build an appropriate model for further behavior analysis. Although some Chinese scholars
have conducted similar research (Shen and Saijo, 2008), their investigations were restricted to a
few large cities. Due to the fact that China is the largest developing country as well as its huge
impact on global environmental protection, our research aims to fill this gap through a national
urban survey.
To assess the current state and nature of environmental concern, and behavior in China today,
this research focused on the expressed opinions about the new environmental paradigm and
related issues and reported environmental behaviors of a general social survey of China in 2003.
It also uses a 2003 urban statistical book, from which our community level variables such as
GDP per capital and pollution indicators are extracted. The following research questions are
addressed:

6

1) How many dimensions should be used to measure environmental concern in China and are
these dimensions different from those in Western societies?
2) How will the selection of particular measure of environmental concern affect its relation to
social-demographic variables?
3) Controlling for individual socio-demographic variables, what is the effect of the
community level variables such as GDP per capita, local air or water pollution level, or industry
structure in the individual‟s community on environmental attitudes and behaviors?
1.4 Dissertation Outline
In addition to this introduction and the summary chapter, this dissertation includes the
following three papers.
Paper one (Chapter 2): The debate on dimensionality of environmental concern still remains.
Inferring from the concept of environmental concern defined by Dunlap and Jones, I attempt to
model measures of environmental concern based on the data collected in 2003 Chinese General
Social Survey through CFA analysis. I assumed that our measurement model includes four latent
factors: balance between economic development and environmental protection, perceived local
environmental problems, pro-environmental behaviors and New Ecological Paradigm. The first
order measurement model has a higher CFI and different strength and direction of association
among those four dimensions than the second order measurement model. Therefore, I argue that
the first order measurement model fits better than its second order measurement model. In
addition, the strength and direction of relations among those four factors indicate that
environmental concern is not a single attitudinal dimension, as some scholars have argued. It also
suggests that modeling environmental concern is complex due to it having multiple dimensions.

7

Paper two (Chapter 3): There is a lot of research that has been done on the social bases of
environmental concern in the Western society. However, we still know little about this in
Chinese society. Employing 2003 Chinese General Social Survey data, I examined the five
hypotheses which are summarized from the literature review through the structural equation
model. Different from results in the Western society, our research indicates that old men have
more environmental concern than young women. People living in the metropolitan areas have
more environmental concern than those living in the small or midsize cities. As expected,
education is positively related to environmental concern; while income has no effects on the
environmental concern. The inconsistent results on the social base of environmental concern
between the Western society and Chinese society deserve further study.
Paper three (Chapter 4): Using 2003 Chinese General Social Survey Data and some selected
variables from Urban Statistical Book in China, I use a Hierarchical Linear and Nonlinear Model
based on the proposed analytic framework will estimate the effects of individual and municipal
level variables on environmental attitudes and two types of environmental behaviors: public vs.
private behaviors. In the individual level, I find that education and gender are two strongest
predictors. Our analysis shows that different from our expectation, some environmental pollution
indicators in municipal level do not play a significant role in environmental attitude and two type
of pro-environment behavior. In addition, some economic indicators in municipal level, such as
GDP per capita are also not significantly related to some type of pro-environmental behavior.
This result challenges our traditional wisdom. This study has also great policy implications for
government policy making. It will help government agencies develop more effective
environmental policies, which lead more and more pro-environmental behaviors.

8

Finally, is the chapter of summary and conclusions. This session outlines some important
research conclusions from those three papers. Through the analysis of CGSS2003 data, I find
that the dimensionality of environmental concern includes four dimensions, which share some
commonalities as well as differences with previous research in the Western society. In the multilevel analysis of environmental attitudes and pro-environmental behaviors, the results indicate
that objective problem hypothesis does not get a full support in the Chinese situation. In the end,
this session discusses their policy implications and data limitations.

9

BIBLIOGRAPHY

10

BIBLIOGRAPHY

China Environmental and Cultural Advance Association. (2008). Chinese Public Environmental
Protection Index
http://www.tt65.net/zhuanti/zhishu/2008gongzhongzhishu/zishuzongshu2008.pdf.
Harris, P. G. (2004). "Getting rich is glorious": Environmental values in the People's Republic of
China. Environmental Values, 13, 145-165.
Harris, P. G. (2006). Environmental Perspectives and Behavior in China - Synopsis and
Bibliography. Environment and Behavior 38(1), 5-21.
Hunter, L. M. (2000). A Comparison of the Environmental Attitudes, Concern, and Behaviors of
Native-born and Foreign- Born U.S. Residents. Population and Environment: A Journal of
Interdisciplinary Studies, 21, 565-580.
Li, H.-l. (1998). Some Thoughts on Confucianism and Ecofeminism. In M. E. Tucker & J.
Berthrong (Eds.), Confucianism and Ecology: The Interrelation of Heaven, Earth, and Humans
(pp. 293-307). Cambridge, MA: Harvard University Press.
Liu, J., & Diamond, J. (2005). China's Environment in a Globalizing World. Nature, 435 11791186.
Marguart-Pyatt, S. T. (2007). Concern for the Environment among General Publics: A Crossnational Study. Society and Natural Resources, 20, 883-898.
Peterson, A. L. (2001). Being Human: Ethics, Environment, and Our Place in the world.
Berkley: University of California Press.
Rumbaugh, T., & Blancher, N. (2004). China: International Trade and WTO Accession, IMF
Working Paper (Http//:imf.org/external/pubs/ft/wp/2004/wp0436.pdf).
Shandra, J. M., London, B., Whooley, O. P., & Williamson, J. B. (2007). International
Nongovernmental Organizations and Carbon Dioxide Emissions in the Developing World: A
Quantitative, Cross-National Analysis. Sociological Inquiry 74(4).
Shapiro, J. (2001). Mao’s War against Nature: Politics and the Environment in Revolutionary
China. Cambridge: Cambridge University Press.
Shen, J., & Saijo, T. (2008). Reexaming the relations between socio-demographic characteristics
and individual environmental concern: Evidence from Shanghai data. Journal of Environmental
Psychology, 28, 42-50.
World Bank and SEPA. (2007). The Cost of Pollution: Economic Estimates of Physical
Damages:
11

http://siteresources.worldbank.org/INTEAPREGTOPENVIRONMENT/Resources/China_Cost_
of_Pollution.pdf.

12

Chapter 2 Environmental Concern in China
2.1 Background
After World War II, fast economic development, population growth and modern technology
change have caused the overuse and destruction of natural resources. As environmental problems
gradually become recognized as pressing social problems, social scientists pay more and more
attention to environmental problems and their related issues. As a result, many surveys have been
used to document the degree to which the public perceived environmental problems as serious
and supported efforts to solve them as well as trends in environmental concern over time (Buttel,
1975). These studies have grown fast in number. By the end of the 1970s, there were about 300
empirical studies of environmental concern by environmental sociologists and other social
scientists from a wide variety of disciplines (Dunlap and Van Liere, 1978). Early efforts to
improve the measures of environmental concern followed two directions (Van Liere and Dunlap
1981). First, researchers increasingly used multiple items rather single item indicators to measure
environmental concern to improve reliability and advanced statistical tools. Second, in order to
make measurement comparisons possible among studies, they tried to standardize measures.
In subsequent years, empirical research on environmental concern has still expanded
enormously. In spite of the burgeoning body of research on environmental concern and its
related topics in industrialized countries, there are three main foci related to these studies on
environmental concern. The first focus is the measure of environmental concern. Due to the
unclear definition of environmental concern, survey design, and the structure of environmental
opinion itself, the dimensionality of environmental concern has been the center of debate
(Dunlap and Jones, 2002; Xiao and Dunlap, 2007; Marquart-Pyatt, 2008; Guber, 1996).This
debate still continues in the Western world.
13

The second focus is to examine the social bases of environmental concern. These studies
attempt to examine variation in concern for environmental quality among differing sectors of the
public. However, difficulties in research on environmental concern might cause inconsistent
findings of social demographic correlates of environmental concern.
The third focus is to study a corresponding relationship between environmental concern and
behaviors based on different theoretical perspectives. This approach assumes that environmental
concern can work as an important predictor of environmental behaviors (Kaiser et al., 1999). For
example, in the Value-Belief-Norm Theory, environmental attitude measured by the New
Ecological Paradigm scale is an important mediated variable in predicting different types of
environmental behaviors (Stern, 2000).
As noted above, how to measure environmental concern plays a key role in determining
research quality of the latter two trends methodologically and theoretically. In methodology,
examining its dimensions is necessary for developing reliable measures; while in theory, its
dimensionality is closely related to meaningful construct of environmental concern (Xiao and
Dunlap, 2007). As several researchers (Van Liere and Dunlap, 1981; Klineberg et al., 1998)
pointed out, that in terms of demographic predictors of environmental concern, how it is
measured makes a difference. In the similar rationale, the measurement of environmental
concern also determines the explanatory power of environmental concern when it works as a
predictor of environmental behaviors.
In the West, environmental concern has been conceptualized in numerous ways. Still, we
know little about environmental concern in developing countries, especially in China. The
Chinese General Social Survey in 2003 provides such a chance to examine and model
environmental concern in urban Chinese residents

14

The present paper proceeds as follows. I will first introduce a brief description of
environmental concern. Then, I will provide a review of literature on dimensionality of
environmental concern. Later, I will offer a detailed description of the analytical strategy, data
and variables. Following this, I will report and interpret key findings. Finally, I will conclude
with a discussion of this research‟s implication and limitation.
2.2 A Brief Introduction of Environmental Concern
As environmental problems become more complex, the meaning of environmental concern
is also more diverse due to the definition of environmental concern related to inclusion of some
environmental problems. Some researchers posit that the meaning of environmental concern is
self evident and doesn‟t need a clear definition, while others assume that environmental concern
is equated with environmental attitudes. Still, various definitions of environmental concern have
been offered by different scholars. Although environmental concern is a broad concept, it is
necessary to clarify the meaning of environmental concern and to make better sense of current
research of its construct.
The earliest definition was provided by Dutch scholars, Nelissen and Scheurs (1975). They
argued that environmental concern represented “the totality of ideas on the protection and control
of and interference with the natural and artificial environment, as well as the behavioral
dispositions connected with them” (Ester, 1981:86). Later, Ester and van deer Meer (1982:72)
defined environmental concern “as the degree to which a person recognizes environmental
problems and is ready to contribute to their solution.” An article by Fransson and Gä
rling
(1999:370) summarized environmental concern as “an evaluation of or an attitude towards facts,
one‟s own behavior, or others‟ behavior with consequences for the environment. Later, in a
recent review paper by Dunlap and Jones (2002:485), environmental concern is referred to as

15

“the degree to which people are aware of problems regarding the environment and support
efforts to solve them and /or indicate a willingness to contribute personally to their
solution.”Although there are many versions in the definitions of environmental concern, I believe
that Dunlap and Jones‟s definition (2002) provide the most comprehensive one and will be used
here as our guiding framework.
Taken these definitions together, Van Liere and Dunlap (1981) pointed out that
environmental concern consists of two components: the “environmental” and “concern”
component. In their view, the environmental component represents substantive content of
environmental concern and is examined by environmental topics chosen from respondents, while
the concern component refers to individual‟s expressions of concern about environmental issues.
As the universe of environmental issue is vast, the inherent complexity of the environmental
component also helps to explain the huge diversity in existing measures of environmental
concern. Since environmental issues take place in different geographical levels, it also gives us a
caution that we need to clarify if they are local or global environmental issues when trying to
measure the environmental component of environmental concern.
In terms of the concern component, there are also two broad approaches to conceptualizing it.
One is a policy approach, in which researchers focus on the studies of public opinion toward
environmental issues. The other is a theoretical approach and its studies conceptualize the
concern component based on the theory of attitude consisting of affective, cognitive and conative
dimensions (Dunlap and Jones, 2002). These two approaches differ in their emphasis on the role
of individuals and their behaviors in creating and solving environmental problems. A theoretical
approach focuses on individuals‟ beliefs, attitudes, intentions and behaviors in the related
environmental behaviors. The policy approach emphasizes the role of social institutions, policies

16

and collective action in protecting the environment. Frequently used dimensions in the policy
approach include perceptions of the seriousness of environmental problems, support for
government policy, and an individual‟s willingness to pay for environmental protection. Since
the data used here does not include items necessary for attitude theory, I will take the policy
approach to measure the concern component of environmental concern.
2.3 Literature on the Dimensions of Environmental Concern
Since the field of environmental concern is relatively new and survey researchers design their
questionnaires independently and statistical tools are changing over time, researchers are inclined
to use different environmental measurements. It is estimated that there are over 1000 published
studies on the measurement of environmental concern (Dunlap and Jones, 2002). Their
measurements of environmental concern range from 1 dimension to 6 dimensions. Therefore, we
can only examine some important and representative studies and hopefully illustrate the
conceptual distinction mentioned above and finally provide a roadmap for understanding existing
measurement efforts.
Earlier studies usually take one dimension as measure of environmental concern. The widely
used one is the New Ecological Paradigm (NEP) scale (Dunlap and Van Liere, 1978; Dunlap et
al., 2000). When environmental problems achieved a prominent position in the mid- to late- 20th
centuries, sociologists began to reflect on the tradition of the discipline of sociology regarding
the environment (Dunlap & Van Liere 1978; Buttel 2002;Dunlap et al. 2002). In 1978, Dunlap
and Van Liere summarized their efforts to measure a fairly new environmental attitude through
the NEP scale with 12 items. Dunlap and Catton (1979) found that there had been a paradigm
shift from the anthropocentric worldview underlying contemporary and classical theories,
labeled the "human exceptionalism paradigm" (HEP), to a "new environmental paradigm"

17

(NEP), which has been renamed in subsequent versions as the "new ecological paradigm". In the
view of Dunlap and Canton, the HEP makes mainstream sociologists not only neglect the
importance of environmental problems, but also accept the optimism inherent in the Dominant
Western Worldview (DWW) by assuming that the endless economic growth will not be
threatened by resource scarcities or other ecological constraints (Dunlap, 2002).
In 2000, Dunlap and his colleagues (2000) revised the NEP scale and made it consist of 15
items. 2003 General Social Survey in China applied 2000 revised NEP scale. Despite its
widespread use, several studies conducted in the U.S have challenged whether the NEP scale is a
single construct or inherently multidimensional. Through using identical measure in a survey of
Iowa residents, Albrecht et al. (1982) concluded that the NEP can be loaded onto three distinct
dimensions. They labeled these three dimensions as “Balance of Nature”, “Limits to Growth”,
and “Human over Nature”. Geller and Lasley (1985) argued that their findings across three
different samples identified 3 dimensions of the NEP but cannot confirm the same factor
structure across samples. Later, through analyzing a survey conducted in Pennsylvania residents,
Scott and Willits (1994) identified the NEP scale as 2 dimensions: “Humans-with-Nature” and
“Balance of Nature/Limits of Growth”.
In China, some scholars also began to focus on the measure of environmental concern. The
NEP scale was introduced by Hong in the 2003 general social survey. Two years later, based on
this data, Hong (2006) pointed out that some items in the NEP scale have a low reliability due to
Chinese social background. Further research using the same data failed to provide the convincing
evidence of a single construct of environmental concern in China (Xiao and Hong, 2007). Using
the data collected from Shanghai in China, Shen and Saijo (2008) identified 3 factors through
exploratory factor analysis (EFA). They labeled these three factors as concern about general and

18

global environmental problems, concern about local environmental problems and proenvironmental behavior measure respectively.
However, most previous studies I mentioned above only include bivariate correlations
analysis and EFA. Although EFA can be a useful tool in detecting possible patterns and
relationships in the data, confirmatory factor analysis (CFA), CFA is superior to EFA in two
aspects. First, researchers can specify a measurement model in advance based on priori
information about the structure of environmental concern. Second, it allows researchers to take
into account both measurement error that is uncorrelated with other variables and measurement
errors that are correlated with each other.
Several studies highlight CFA‟s potential for helping us understand the measurement of
environmental concern. The first study was conducted by Guber (1996). After pointing out
disadvantages in previous research, she used national Gallup surveys to develop and test a model
of environmental concern. She proposed three latent factors: perceived seriousness of
environmental

problems,

pro-environmental

behaviors

and

self-identification

as

an

environmentalist. She found that these three dimensions are highly correlated ranging from 0.64
to 0.8, which led her conclude that environmental concern is a single construct. Later, Guber
(2003) revised her model by “replacing pro-environmental behavior with an “environmental
policy preference” dimension. Unfortunately, her study also has some pitfalls as Dunlap and
Jones (2002) pointed out. First, she did not make a clear distinction between global and other
environmental problems. Second, in the concern component of environmental concern, she
ignored items that deal with tradeoff between economic development and environmental
protection. Third, its measure of self-identification only included a single item indicator, which
might cause measurement quality problems.

19

Two years later, Carman (1998) used Michigan‟s National Election Survey data to measure
environmental policy support. Carman contended that environmental policy support in the
United States consists of three latent factors: the environmental economic concern factor, which
reflects a willingness to give environmental protection priority over economic growth, the
environmental regulation concern factor reflecting support for such regulations, and the
environmental quality assessment factor reflecting perceived seriousness of environmental
problem. Carman‟s finding demonstrated that environmental policy support in the United States
is “a complex, multidimensional, and multilevel set of attitudes.” (Carman, 1998:731) However,
like Guber‟s research, Carman‟s research also has some weaknesses. First, he did not make a
clear distinction between environmental policy support and environmental concern. Second,
Dunlap and Jones (2002:514) argued, he “failed to see the possibility that his measure of
perceived seriousness may not correlate as highly as the economic and regulatory dimensions.”
Third, Carmen also did not consider the geographical levels when he evaluated perceived
seriousness of environmental problems.
In the same year, using the Texas Environmental Survey, Klineberg et al. (1998) proposed
four measures of environmental concern: (1) economic/government trade-offs ,(2) perceived
seriousness of pollution,(3) reported pro-environmental behaviors, and (4) ecological worldview.
Each measure consists of multiple items. Their findings suggest that the determinants of
environmental concern vary in a predictable way and that it really does make a difference how
environmental concern is measured. However, there are also some pitfalls in this research. First,
they did not explore relationships among these four measures of environmental concern. Second,
they used a summary index to represent each measure of environmental concern and failed to
verify each summary index‟s internal consistency. Third, like Carmen‟s model, they used

20

perceived seriousness of pollution as one dimension, and did not make a clarification if such
pollution is perceived in the global level or local level.
The most recent CFA study on environmental concern by Xiao and Dunlap (2007) employs a
comparative design by using national probability samples of citizens from Canada and the
United States. They identified eight facets of environmental concern: community environmental
issues (C-ISSUES), national environmental issues (N-ISSUES), global environmental issues (GISSUES), perceived importance of environmental problems (IMPORTANCE), willingness to
pay for environmental protection (ENV-ECO-TRADEOFFS), support for governmental policies
and regulations (POLICY SUPPORT), new ecological paradigm (NEP) and pro-environmental
behaviors and activism (BEHAVIOR &ACTIVISM). Their final model indicates that except for
N-ISSUES and C-ISSUES, six other key facets of environmental concern have moderate to high
factor loadings on one underlying construct, environmental concern. The similarity in the U.S.
and Canadian findings increases the confidence in the validity of their measurement model of
environmental concern. Unfortunately, their measurement model also has some disadvantages.
First, it tried to establish a comprehensive model with a second order latent construct while they
ignored some important factors such as, C-ISSUES and N-ISSUE in their final measurement
model. Second, different pro-environmental behaviors have different social-demographic
characteristics and causes (Stern et al., 1999). It is inappropriate that these different types of
environmental behaviors can be regarded as a general type in the measurement model.
In a word, environmental concern has different measures ranging from 1 dimension to 6
dimensions. However, most of this research focusing on environmental concern was conducted
in advanced industrial countries such as the United States (Dunlap et al., 2000; Klineberg et al.,
1998), and their measurement model might be also limited to those societies. By comparing

21

industrialized countries, Marquart-Pyatt (2008) also pointed out that difference remains in terms
of measuring environmental concern in the industrialized countries. Even if a comprehensive
conceptualization model was claimed by Xiao and Dunlap (2007), they also acknowledged that
one should be cautious to apply their measurement model outside U.S and Canadian societies.
Environmental concern is thought to be the result of a complex interaction of many social,
economic, cultural and regional factors. Knowing that there are huge differences between China
and most western countries in social, economic, cultural, and political systems, it is worthwhile
and necessary to build a measurement model which is fit with the Chinese society. The present
study aims to fill this gap, because China is the largest developing nation and its environmental
impact is also the largest. I believe that this study based on the Chinese survey will help to build
a measurement model which is suited for the Chinese society background and examine internal
relationships among those measurements and then find whether measures of environmental
concern in China are different from those in industrialized societies.
2.4 Data and Method
The dataset we use in this paper comes from the General Social Survey of 2003 administered
jointly by the Department of Sociology at Renmin University of China and the Survey Research
Center of the Hong Kong University of Science and Technology. The survey was conducted in
urban regions in the mainland of China, and respondents were sampled in five stages. The first
stage targeted districts of cities and counties of 22 provinces, 3 autonomous regions (provinciallevel administrative subdivisions of China), and 4 central government designed municipalities
(the most economically and politically important cities among the highest populations of Chinese
cities, major cities hereafter). Tibet, an autonomous region, was excluded due to logistic
constraints. The second through fifth stages differed by targeting townships of counties or sub-

22

districts of cities, neighborhood committees in towns or city sub-districts, households, and
eligible respondents respectively.
The data was collected by face to face interviews in the randomly selected urban areas in
China. Based on the sample frames, the name list of the household in each selected area was
prepared through various channels like the household registration institutions and residential
committees. The interviewers were trained both at the national and provincial levels through
lecturing, imitated interviews, discussions and workshops. To improve the responses rate, each
interviewer is provided with the formal introduction letter issued by Renmin University of China.
A formal introduction letter was used to establish some kind of mutual trust between letter
holders and interviewees. The survey was warmly welcomed and actively coordinated by the
interviewees and their families through giving selected household gifts and other material
benefits.
The questionnaire includes two parts : Questionnaire A and measurement of environmental
items (Questionnaire B). Due to the difficulty of obtaining cooperation from local partners,
Questionnaire B was not conducted in

four provinces (Heilongjiang, Jilin, Hubei, and

Guangdong), which caused the current sample to be 902 respondents less than originally
designed. The random sample included 5980 people, and 5894 respondents were successfully
interviewed (98.6% of the sample). For those returning both questionnaires A and B, the sample
size is 5073 (Hong and Xiao 2007) ,whose respondents are living in 88 city districts. Survey data
were collected through in-person interviews. In terms of some missing values, I use mean value
to replace them. In the Questionnaire B, there are 48.2% male respondents and 51.8% female.
The respondents‟ average age is 43 years and average education level is high school (12 years).

23

Questionnaire B is the first survey that research agencies conducted to investigate Chinese
environmental attitudes and behaviors on the national scale.
Different from previous studies relying on exploratory factor analysis or principal component
analysis, confirmatory factor analysis (CFA) will be applied in this paper. CFA is superior in that
it tests latent sources of variation and covariation based on prior specified patterns or
relationships among the dimensions of environmental concern (Guber, 1996). In addition to its
great emphasis on theory testing, the CFA framework can offer many other analytic tools, which
are not available in EFA (Brown, 2006).
2.5 Measurement Model of Environmental Concern
To investigate the dimensions of environmental concern, it is necessary to follow Dunlap and
Jones‟ definition (2002) and previous studies based on CFA (Guber, 1996, 2003; Carmen, 1998;
Klineberg et al. 1998; Xiao and Dunlap 2007). I assume that environmental concern is a latent
construct with four factors and each factor has multiple indicators. I propose a latent construct of
environmental concern as consisting of four dimensions: (1) environmental-economic tradeoffs
(Econ Trade Off), (2) pro-environmental behaviors (PEB), (3) perceived local environmental
problems (Perceived Local Problems) and (4) new ecological paradigm (NEP). Local
environmental problems can be considered as “the degree to which people are aware of local or
global problems, respectively regarding the environment” Environmental-economic tradeoffs
and pro-environmental behaviors represents “support for efforts to solve them and/or indicate a
willingness to contribute personally to their solution” in Dunlap and Jones‟ definition of
environmental concern.
The environmental-economic tradeoffs dimension is essentially the same as economic
concern factor in Carmen‟s (1998) model, which include four indicators: environmental spending,

24

increasing car cost, protecting environment and maintaining jobs, and restricting business. Xiao
and Dunlap (2007) also included this factor in their model and this factor consists of four
indicators which range from the priority to economy vs. environment to the priority of job
security vs. preventing a polluted environment However, these four indicators have different
scales. It might influence the validity of measurement. There are two types of validity: construct
validity and predictive validity. Since these indicators have different scales, it might confuse
respondents and then reduce construct validity.
Both Guber‟s model and Carmen‟s model did not make a geographical classification of
perceived environmental problems factor. In Guber‟s model, the perceived seriousness of
environmental problems factor only focuses on global environmental problems. The qualitative
assessment factor in Carmen‟s model included two national level environmental problems and
two local level environmental problems. In the model tested by Xiao and Dunlap (2007), they
included beliefs about three levels of perceived environmental problems: global, national, and
community level. After a series of CFAs, they just kept globally perceived environmental
problems in their final model. However, in my model, due to data limitations, I focus on
perceived local level environmental problems.
Guber‟s model and Xiao and Dunlap‟s model include a pro-environmental behavior factor,
but they used different items to measure pro-environmental behaviors. For example, the
behaviors and activism factor in Xiao and Dunlap‟s model includes three types of environmental
behaviors.

This causes some indicators of pro-environmental behavior to have lower item

reliabilities than other measurement indicators. Here, based on a series of preliminary analyses, I
construct pro-environmental behaviors with high item reliabilities.

25

Like the model constructed by Klineberg et al. (1999) and Xiao et al. (2007), I added the
NEP scale as one dimension of environmental concern, because it has not only been widely used
to explore the interrelationships between human activities and natural systems but also it has
been frequently used as a general environmental attitude or ecological worldview. Each of these
dimensions is measured with multiple items. Of these four dimensions, they capture policyrelevant facets of environmental concern and are measured by eighteen items in total.
I hypothesized that there are intercorrelations among those four dimensions as I describe
below.
H1. The perception of local environmental problems has a positive relationship with
ecological worldview measured by the NEP. That is, those people who have perceptions of more
serious local environmental problems can transform their traditional worldview to ecological
world view easily.
H2. The perception of local environmental problems has positive association with proenvironmental behaviors. Those people who have perceptions of more serious local
environmental problems are more willing to take pro-environmental behaviors.
H3. The NEP has a positive correlation relationship with environmental-economic tradeoff
factor.-The formation of their ecological worldview can be expected to help people make some
adjustments in public policies of environmental protection and then reach a balance between
environmental protection and economic growth.
H4. The economic trade off factor has positive correlation relationship with perceived local
environmental problems. Those people who have more perceptions of seriously local
environmental problems are more likely to prefer environmental protection to economic growth.

26

Before determining the final measurement model of environmental concern, I conducted a
series of preliminary CFA analyses. Tabachnic and Fidell (2007) suggest that the standardized
loadings under 0.32 indicate that items do not relate strongly enough to the factor and cannot be
used as indicators of the model. This criterion is usually applied to first-order factor loadings.
Here, I use a stricter criterion that the standardized loadings below 0.40 will not be kept in this
model. Therefore, the final measurement model for environmental concern in urban China will
be the Figure 2.1.
All rectangular boxes are indicators, which are questionnaire items, loading onto the factors
showed in ovals. These four factors in ovals are assumed to be correlated with each other. Each
indicator is also assumed to be affected by a measurement error from e1 to e18 (whose names
start with the letter, e).

27

Figure 2.1 Measurement Model of Environmental Concern

1
1

item1

e1
1

Econ trade
off

item2

e2
1

item3
1

e3
1

item4

e4
1

item5

e5
1

Perceived
local problem

item6

e6
1

item7

e7
1

item8

e8
1

item9

e9
1

1

item10

e10
1

PEB

item11

e11
1

item12

e12
1

item13

e13
1

1

item14

e14
1

item15

e15
1

NEP

item16

e16
1

item17

e17
1

item18

e18

Economic trade off factor includes 2 items. The first one is “If a certain place faces the
dilemma of developing economy and protecting environment, which of the following view
answers for your opinion better?” There are three choices for this item: 1) development first; 2)
Not sure; 3) environment first. The second one is “Suppose in a certain community there is a
28

factory that enjoys economic success and brings many economic benefits to the local residents,
but the factory also produces wastewater that pollutes the riverhead of downstream areas. If you
were one of the residents in the community where the factory rests, which of the following
attitude answer stands for yours better? ” There are also three answers for this question: 1) agree
to close factory; 2) disagree to close factory 3) Not sure.
The factor of perceived local problems originally consisted of 10 indicators. However, I kept
only 7 out of 10 indicators since 3 indicators have standardized factor loadings which are less
than 0.5. The question preceding all of them is “could you tell us how severe these problems are,
which exist in your community?” These 7 indicators include air pollution, water pollution, noise
pollution, industrial waste pollution, urban garbage pollution, green space deficiency, and
destruction of natural vegetation in forests respectively. Answers to these questions are following:
1) very serious; 2) somewhat serious; 3) passable; 4) not very serious, and 5) not serious at all.
The pro-environmental behaviors (PEB) factor is also a latent construct including 3 items.
The survey lists 10 pro-environmental behaviors. CFA analysis showed that there are lower
reliabilities in the measurement of pro-environmental behaviors. Therefore, I kept three items
with standardized factor loading more than 0.5. The statement is “In the past one year, have you
conducted these following behaviors?” The three pro-environmental behaviors include: 1,
donating money for environmental protection; 2, actively participate in environmental campaigns
sponsored by government and workplace; 3. actively participate in environmental protection
activities held by Non-governmental organization. Possible answers to these behaviors are: 1)
Never, 2) Sometimes, 3) Often.
The NEP scale includes 15 indicators. Due to low factor loadings in some items, I only kept
6 out of 15 indicators. Here, I kept factor loadings which are greater than or equal to 0.5 in the

29

preliminary analysis. This is a stricter requirement on factor loadings than Xiao and Hong
(2010)‟s work.
In their paper, they used ten NEP items which included some factor loadings less than 0.4. So
the NEP factor here includes 6 items. The first indicator is “When humans interfere with nature it
often produces disastrous consequences.” The second indicator is “Humans are severely abusing
the environment.” The third indicator is “Despite our special abilities, humans are still subject to
the law of nature.” The fourth indicator is “The earth is like a spaceship with very limited room
and resources.” The fifth indicator is “The balance of nature is very delicate and easily upset.”
The sixth indicator is “If things continue on their present course, we will soon experience a
major catastrophe.” The answers to these indicators are following: 1) strongly agree; 2) mildly
agree; 3) not sure; 4) mildly disagree and 5) strongly disagree.
All indicators are recoded so that higher scores reflect higher levels of environmental
concern. For coding and exact question wording of all items corresponding to each factor, see
Table 2-1.

Table 2-1 Description of Indicators
Factor

Environmental
economic
Trade-Off

Indictor
Item1

Description of indicators
If a certain place faces the dilemma of
developing economy and protecting
environment, which of the following view
answers for your opinion better?

Item2

Suppose in a certain community there is a
factory that brings many economic benefits to
the local residents, but the factory also
produces waste water that pollutes the
riverhead of downstream areas. If you were one
of the residents in the community which of the
following attitude answer stands for your
better?
30

Code
1,Development
first;
2,Not Sure
3,Environment,
first;
1, Disagree to
close factory;
2, Not Sure
3, Agree to
close factory

Table 2-1 (cont’d) Description of Indicators
Factor
Indicator Description of indicators
Item3
Air pollution
Item4
Item5

Noise pollution

Item6

Industrial waste pollution

Item7

Urban garbage pollution

Item8

Green space deficiency

Item9

Perceived
Local
Problems

Water pollution

Destruction of natural vegetation in forests

ProItem10
Environmental
Behaviors
Item11

Item12

Item13
Item14
Item15
NEP

Item16
Item17
Item18

Donating money for environmental protection

Actively
participate
in
environmental
campaigns sponsored by government and
workplace
Actively
participate
in
environmental
protection activities held by Non-governmental
organization
When humans interfere with nature it
often produces disastrous consequences
Humans are severely abusing the environment
Despite our special abilities humans
are still subject to the law of nature
The earth is like a spaceship with very limited
room and resources
The balance of nature is very delicate and
easily Upset
If things continue on their present course,
we will soon experience a major
ecological catastrophe

Code
5-Very serious to
1 -not serious at all
5-Very serious to
1 -not serious at all
5-Very serious to
1 -not serious at all
5-Very serious to
1-not serious at all
5-Very serious to
1 -not serious at all
5-Very serious to
1 -not serious at all
5-Very serious to
1 -not serious at all
3-Often
2-Sometimes
1-Never
3- Often
2-Sometimes
1-Never
3- Often
2-Sometimes
1-Never
5-strongly agree to
1-strongly disagree
5-strongly agree to
1-strongly disagree
5-strongly agree to
1-strongly disagree
5-strongly agree to
1-strongly disagree
5-strongly agree to
1-strongly disagree
5-strongly agree to
1-strongly disagree

2.6 Result and Discussion
Through the CFA analysis, I got the following results (Table 2-2) for the measurement model.
The Chi- square test of this model is statistically significant with the value of 1143.94 and 129
31

degrees of freedom. However, using the Chi-square value to evaluate the model‟s overall
goodness-of-fit is not valid due to the following reasons. First, in many cases, the underlying
distribution of the model is not a Chi-square distribution. Second, the Chi-square value is
influenced by sample size and we have a large sample size. Therefore, in addition to providing
Chi-square statistics, I also report two other measurements of goodness-of-fit: comparative fit
indices (CFI) and root mean square error of approximation (RMSEA). Following Brown‟s
recommendation (2006), when CFI values are close to 0.95 or greater and RMSEA values are
close to 0.06 or below, it indicates this model is a good model fit. In the model I mentioned
above, CFI is 0.943 and RMSEA is 0.039, this suggests that our model with four factors has a
good fit. Each indicator‟s unstandardized and standardized loadings can be found in Table 2-2.
There are five columns in the Table 2-2, the first column includes four factors which I use
to build the measurement model. The second column lists our 18 measurement indicators. The
third and fourth columns are the loadings and standard error. The final one is the t-ratio to test
the significance of the factor loadings.
Table 2-2 Unstandardized (Standardized) Factor Loading from CFA
Factor

Items

Econ
Trade off

Item1
Item2
Item3
Item4
Item5
Item6
Item7
Item8
Item9
Item10
Item11
Item12

Perceived
Local
Problem

ProEnvironmental
Behaviors

Factor
Loadings
1(0.56)
0.85 (0.60)
1(0.72)
0.94(0.68)
0.82(0.60)
0.84(0.63)
0.82(0.58)
0.56(0.43)
0.45(0.42)
1(0.52)
1. 88(0.76)
1. 51(0.76)
32

Standard
Error

t -ratio

0.079

10.75

0.023
0.022
0.022
0.023
0.021
0.017

40.84
36.73
38.33
35.97
26.87
26.30

0.021
0.017

29.56
29.57

Table 2-2 (Cont’d) Unstandardized (Standardized) Factor Loading from CFA
Factor
Items
Factor
Standard
Loadings
Error
1(0.58)
Item13
1.13(0.59)
0.038
Item14
0.74(0.47)
0.03
Item15
NEP
0.89(0.53)
0.033
Item16
0.96(0.57)
0.033
Item17
1.12(0.63)
0.037
Item18
DF
129
Chi-square
1143.94
CFI
RMSEA
0.037
Note: In the column 3, the values in parenthesis are standardized solutions.

t-ratio

29.58
24.98
27.21
28.88
30.51
P<0.001
0.947

From the table above, all the factor loadings are statistically significant at p=0.05 and their
values in standardized form range from 0.42 to 0.76. All the loadings to these four factors are
positive and each value is higher than 0.4, which suggests that the measurement model of
environmental concern in urban China has a high degree of consistency. In all four factors, it is
striking to find that the factor loadings of pro-environmental behaviors have the highest value in
the standardized form and its corresponding factor loading for item 10, item11 and item 12 is
0.51, 0.76 and 0.76 respectively while the factor loading of perceived local problems has a
lowest value and the standardized factor loadings for item9 is 0.42.
Generally speaking, our measurement model has a good fit and each item can be regarded as
a reliable indicator of its factor based on the size of the loading. Although Table 2-2 offers
goodness of fit statistics and its unstandardized and standardized solutions, we still do not know
what these factors‟ internal relationships are? In the next step, I would like to further explore the
correlation among these four factors and their statistical significance. Table 2-3 shows the
correlation among four factors.

33

Table 2-3 Correlation among these Four Factors
Correlation
EconCoefficient
Trade off
Econ Trade off
1
Perceived Local Problems
PEB
NEP
Note ：*p<=0.05, **p<=0.01, ***p<=0.001

Perceived
Local
Problem
-0.026
1

PEB
0.142***
0.114***
1

NEP
0.322***
0.142***
0.149***
1

From table 2-3, all correlations between factors are significant at p < .001 except for
correlation between Econ Trade off factor and Perceived Local Problem factor. Therefore,
Hypotheses 1, 2, and 3 get supports here and only hypothesis 4 cannot be verified in this Table 23. Different from our assumption (H4) in the measurement model, I find that perceived local
problems factor has a near zero and non-significant associated with economic trade off factor.
Therefore, the relationship between perceived local problems factor and economic trade off
factor might be independent.
According to the study by Xiao and Dunlap (2007), due to the small loading value of
perceived local environmental problems, they did not keep this factor in their measurement
model. However, according to definition of environmental concern by Dunlap and Jones (2002)
and deHaven-Smith‟s (1991) conclusion that average citizens‟ environmental concern is largely
localized, it is necessary to keep perceived local problem in the measurement model. In addition,
deleting this factor in our model, the value of CFI and RMSEA is almost the same, so I keep this
factor in this model.
There are two possible explanations for the very weak or even independent relationship
between perceived local problem factor and economic trade off factor. First, China is a
centralized country or authoritarian state. This means that the central government controls and
allocates most of the important resources and the top officials in the local government were
34

appointed by central government. The Chinese government at different levels takes a top -down
policy making process and rules out the possibility of public participation. Without appropriate
participation channels, the perception of local environmental problems by the local residents
cannot be converted to their action in participating the policy making in environmental
protection.
This is quite different from the western societies. For example, in the U.S., if the public
perceives environmental problems in their community, they can launch social movements or
organize social protests to express their opinion. Considering the real situation of China, such
movements or protests are very rare.
Second, according to Inglehart (1995), people in developing countries are more likely to have
materialist values, which pursue economic growth and national security while ignore
environmental quality, than those in developed countries. Based on the same dataset, Hong (2005)
concluded that Chinese people are still in the phase of materialist values. In addition, central
government and Communist party policies also emphasize the priority of economic growth over
other issues. For most of common Chinese people, getting rich is glorious whether such process
will destroy the environment or not. As Chinese people care more about economic growth
instead of environmental quality, it is also understandable that although Chinese people
perceived more local environmental problems while still try to get the benefits from economic
growth and leave consequences of environmental pollution to the next generation
The NEP factor has the strongest positive association with the factor of economical trade off
factor. This suggests that the NEP factor is an important measurement tool for environmental
concern. It also indicates that individuals with a high score in the NEP scale are more inclined to
take environmental protection priority over economic growth. Like other research in western

35

societies, we find that there are positive and significant association between the NEP factor and
pro-environmental behaviors factor. This also reveals the facts that people with the higher NEP
factor score are more likely to conduct pro-environmental behaviors.
It is worth noting that the pro-environmental behaviors factor has a comparatively mild
positive relationship with pro-perceived local problems factor and economic trade off factor.
This conclusion is consistent to conventional thinking.

In the conventional wisdom, local

environmental problems influence local residents more directly than global environmental
problem, which might cause local people to take actions on reducing the influence of local
environmental problems and those who conducted more pro-environmental behaviors are more
likely to favor environmental protections prior to economic growth.
To simplify our measurement model, can a high order measurement model with these four
factors represent a more meaningful construct of environmental concern than our measurement
model above? The second order factor will be environmental concern while the first order factors
still keep the same items. However, through data analysis, I prefer not to take a second order
measurement model based on the following reasons. First, from goodness of fit indices, CFI and
RMSEA in the second order measurement is 0.94 and 0.04 (CFI is 0.943 and RMSEA is 0.39 in
the first order measurement model). Therefore, the first order measurement model seems to fit
my data essentially equal to high order measurement. Second, perceived local environmental
problem factor has a different correlation pattern with other factors. It is unwise to ignore
complex correlation relationship among those four factors just for making a model parsimonious.
Third, a single second order model‟s factors have relatively equal correlation strength and the
same direction. However, the correlation matrix among those four factors in our measurement

36

model ranges from -0.026 to 0.322. Thus, I conclude that my measurement model represents a
more meaningful construct of environmental concern.
2.7 Conclusion
In this study, adopting Dunlap and Jones‟s conceptualization of environmental concern and
focusing on its policy orientation, I followed the example of Guber (1996), Carman (1998),
Klineberg et al. (1998) and Xiao and Dunlap (2007) in applying CFA to avoid the traditional
shortcomings of more commonly utilized techniques such as EFA and PCA to evaluate the
dimensionality of environmental concern.
I used the 2003 Chinese general social survey data to address an important question: is the
measurement of environmental concern in China different from the measurement of this concept
in Western societies? In terms of the dimensions measured between the West and China, My
answer is that measures of environmental concern in China share some commonalities with those
measures in Western societies; however, it also has some distinctions. The four dimensions I
chose here can be found in the most of Western research. For example, Guber‟s (1996) model
includes perceived seriousness of environmental problems factor and pro-environmental
behavior factor while Xiao and Dunlap model (2007) consists of environmental economic
tradeoff factor and the NEP factors. Even though our measurement model shares two factors
with Guber‟s (1996) model, Guber used additive indexes to measure these two factors.
Compared to Carman‟s model (1998), our measurement model is not limited to the dimension of
support for environmental policies.
The results indicate that environmental concern in China includes four dimensions:
economical trade off, pro-environmental behavior, perceived local problems and NEP. In
general, these measures are well represented as multiple item latent constructs in the national

37

sample. The measurement model I proposed is different from Xiao and Dunlap‟s second order
measurement model based on Western societies, which was regarded as the most comprehensive
measurement model in current Western study, while ignored the internal correlation relationship
among all six key dimensions of environmental concern. Here, I explore the correlation pattern
among the constructed four latent factors and found that this pattern has some kinds of specialty
rooted in the Chinese society. For example, the independent relationship between perceived
local problems factor and economic tradeoff factor might reveal the fact that unlike the common
people in western societies, most of Chinese people lack channels in participating policy decision
making. In addition, in China the correlation among our four factors are less than 0.2, while in
Guber‟s (1996) study, the correlations among three factors are quite high at 0.64, 0.73 and 0.8 in
the U.S.
My analysis indicates that the NEP factor has a high association with economical trade off
factor and pro-environmental behavior factor, which is consistent with the concept of
environmental concern developed by Dunlap and Jones (2002). It emphasizes its policy approach
since it focuses on the degree of willing to take actions to environmental protection in individual
level and collective level.
Although some previous studies recognized the importance of perceived local environmental
problem in constructing environmental concern (Xiao and Dunlap, 2007), their research did not
take perceived local environmental problem into account in their constructed model. Through
adding perceived local environmental problems factor in my measurement model, I found that
except for the relationship between economic trade off factor and perceived local environmental
problems factor, other relationships among these factors are fit with the original expectation.,
Perceived local environmental problems factor have no statistically significant correlation with

38

the economical trade off factor. Does it mean that the distinctiveness of the perceived seriousness
of local environmental problem only exist in China due to its social, economical, and political
system? This question is in need of further research.
To my knowledge, this research is the first systematic study of environmental concern with
CFA in China, a country with the largest population in the world where we know little about the
level of environmental concern and how to measure it. This model can be used as a baseline for
the future research based on the Chinese society background. This research also has implications
to establish a corresponding link between attitudes and behaviors. However, since our sample
does not include rural Chinese residents, it cannot be regarded as a representative of all Chinese.
This also restricts measurement model of environmental concern to urban residents in China.
Future research should focus on if there is rural and urban difference in the measurement of
environmental concern.

39

BIBLIOGRAPHY

40

BIBLIOGRAPHY

Albrecht, D., Bultena, G., Hoiberg, E., & Nowak, P. (1982). The NewEnvironmental Paradigm
Scale. Journal of Environmental Education, 13, 39-43.
Brown, T. A. (2006). Confirmatory Factor Analysis for Applied Research. New York: The
Guilford Press.
Buttel, F. H. (1975). The Environmental Movement: Consensus, Conflict and Change. Journal of
Environmental Education, 7, 53-63.
Buttel, F. H. (2002). Environmental Sociology and the Classical Sociological Tradition: Some
Observations on Current Controversities. In R. E. Dunlap, F. H. Buttlel, P. Dickens & A.
Gijswijt (Eds.), Sociological Theory and the Environment: Classical Foundations,Contemporary
Insights. New York: Rowman&Littlefield Publishers, INC.
Carman, C. (1998). Dimensions of Environmental Policy Support in the United States. Social
Science Quarterly, 79, 717-733.
deHaven-Smith, L. (1991). Environmental Concern in Florida and the Nation. Gainesville, FL:
University of Florida Press.
Dunlap, R. E. (2002). Paradigms, Theories, and Environmental Sociology. In R. E. Dunlap, F. H.
Buttel, P. Dickens & A. Gijswijt (Eds.), Sociological Theory and the Environment: Classical
Foundations, Contemporary Insights. New York: Rowan&Littlefield Publishers, INC.
Dunlap, R. E., & Catton, W. R. J. (1979). Environmental Sociology. Annual Review of
Sociology, 5, 243-273.
Dunlap, R. E., & Jones, R. E. (2002). Environmental Concern: Conceptual and Measurement
Issues. In R. E. Dunlap & W. Michelson (Eds.), Handbook of Environmental Sociology (pp. 482524). Westport, CT: Greenwood Press.
Dunlap, R. E., & Valn Liere, K. D. (1978). The "new environmental paradigm": A Proposed
Measuring Instrument and Preliminary results. Journal of Environmental Education, 9, 10-19.
Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). Measuring Endorsement
of the New Ecological Paradigm: A Revised NEP Scale. Journal of Social Issues, 56(3), 425442.
Ester, P. (1981). Environmental Concern in the Netherlands. In T. O'Riordan & R. K. Turner
(Eds.), (Vol. 3, pp. 81-108). Chichester: John Wiley & Sons.
Ester, P., & Meer, F. v. d. (1982). Determinants of Individual Environmental Behavior: An
Outline of a Behavior Model and Some Research Findings. The Netherlands' Journal of
41

Sociology, 18, 57-94.
Fransson, N., & Gä
rling , T. (1999). Environmental Concern: Conceptual Definitions,
Measurement Methods and Research Findings. Journal of Environmental Psychology, 19, 369382.
Geller, J. M., & Lasley, P. (1985). The New Environmental Paradigm Scale: A Reexamination.
Journal of Environmental Education, 17, 9-12.
Guber, D. L. (1996). Environmental Concern and the Dimensionality Problem. Social Science
Quarterly, 77, 644-662.
Guber, D. L. (2003). The Grassroots of a Green Revolution, Polling America on the
Environment. Cambridge, MA: MIT press.
Hong, D. (2005). Urban Chinese Residents' Environmental Concern Jiangsu Social Science,
1(127-132).
Hong, D. (2006). Measurement of Environmental Concern: NEP Scale's Application in China.
Chinese Journal of Society, 5, 71-91.
Inglehart, R. (1995). Public Support for Environmental Protection: Objective Problems and
Subjective Values in 43 Societies. PS: Political Science and Politics, 15.
Kaiser, F. G., Sybille, W., & Fuhrer, U. (1999). Environmental Attitude and Ecological
Behavior. Journal of Environmental Psychology, 19, 1-19.
Klineberg, S. L., McKeever, M., & Rothenbach, B. (1998). Demographic Predictors of
Environmental Concern: It Does Make a Difference How it's Measured. Social Science
Quarterly, 79(734-753).
Marquart-Pyatt, S. T. (2008). Are There Similar sources of environmental concern? Comparing
Industrialized countries. Social Science Quarterly, 89(5), 1312-1335.
Nelissen, NL.M. and Schreurs, L.(1975),. Het Meten van Milieubesef, Maastricht
Scott, D., & Willits, F. K. (1994). Environmental Attitudes and Behaviors: A Pennsylvania
Survey. Environment and Behavior, 26(2), 239-260.
Shen, J., & Saijo, T. (2008). Reexaming the relations between socio-demographic characteristics
and individual environmental concern: Evidence from Shanghai data. Journal of Environmental
Psychology, 28, 42-50.
Stern, P. C. (2000). Toward a Coherent Theory of Environmentally Significant Behavior.
Journal of Social Issues, 56, 407-424.

42

Stern, P. C., Dietz, T., Abel, T., Guagnano, G. A., & Kalof, L. (1999). A Value Belief- Norm
Theory of Support for Social Movements: The Case of Environmental Concern. Human Ecology
Review, 6, 81-97.
Tabachnic, B. G., & Fidell, L. S. (2007). Using Multivariate Statistics (5th Ed.). . Boston,MA
Allyn and Bacon.
Van Liere, K. D., & Dunlap, R. E. (1981). Environmental Concern: Does It Make a Difference
How It's Measured. Environment and Behavior, 13, 651-676.
Xiao, C., & Dunlap, R. E. (2007). Validating a comprehensive Model of Environmental Concern
Cross-Nationally: A U.S.-Canadian Comparison. Social Science Quarterly, 88(2), 471-493.

43

Chapter 3 Social Demographic Characteristics of
Environmental Concern
3.1 Introduction
Over the past three decades, interests in issues of environmental concern have provoked a
huge amount of research on the social-demographic predictors of environmental concern.
Growth in the number of these studies has been accompanied by focusing on the measurement of
environmental concern and its social bases. Although there are difficulties inherent in gauging
this complex concept, two conclusions can be drawn from those studies related to its social
determinants. First, when all social demographic predictors are combined, they explain a small
portion of the variance in environmental concern (Van Liere and Dunlap, 1980, Buttel and Flinn,
1978, Jones and Dunlap, 1992, Fransson, 1999).

Second, in terms of social demographic

variable, there are no consistent results in predicting environmental concern (Van Liere and
Dunlap, 1980; Klineberg et al., 1998; Mohai et al., 2010; Mobley et al., 2010).
The findings above are based on the background of Western societies, though some
researchers argue that the model specification errors associated with the inherent assumptions of
the statistical tools partially caused it (Xiao and McCright, 2007). According to the view of Van
Liere and Dunlap (1981:653), existing measures of environmental concern differ in two aspects:
the “substantive issues” included in the measure, and the “theoretical conceptualization” used in
developing the measurement. As a result, various combinations of substantive issues and
theoretical conceptualization have caused many different measures. For example, in terms of
substantive issues, some researchers (Tognacci et al. 1972; Lounsbury and Tornatzky, 1977)
have treated attitudes toward pollution, population, and natural resources as distinct dimensions
of environmental concern, while other researchers tended to combine these substantive items into
a single environmental concern measure (Maloney et al., 1975; Weigel and Weigel, 1978; Buttel
44

and Flinn, 1976). As for the debate of theoretical conceptualization of environmental concern,
this has implications for what specific components should be included to measure environmental
concern (Van Liere and Dunlap 1981). For example, can environmental problems perceived by
individuals be classified into different geographic levels according to those problems‟ sphere of
influence? Should tradeoffs between economic growth and environmental protection be
employed in the measure of environmental concern?
Even with the same measurement scale, researchers are still diverging as to whether this
scale can be divided into several dimensions or subscales. For example, Dunlap and Van Liere
(1978, 2000) proposed the new ecological paradigm (NEP) scale to measure public environmental
concern and argued that the scale is both a valid and reliable indicator of it. However, Albrecht et
al. (1982) concluded that the NEP scale can be divided into three distinct dimensions which are
labeled respectively as “Balance of Nature”, “Limits to Growth”, and “Human over Nature”. Still
today, many scholars use this scale as a general measure of environmental concern (Hong, 2006,
Xiao and Hong, 2010; Dunlap and Van Liere, 2000). Therefore, to understand the relationship
between social demographic variables and environmental concern, specifying which measures of
environment concern we are examining is extremely necessary.
The purpose of the present study is to empirically examine social demographic predictors of
various environmental concern measures. First, I review related literature and explore their
variation of social demographic variables in predicting environmental concern both in Western
Society and in China. Then, I outline four distinct ways of measuring that concern: 1) tradeoff
between economic growth and environmental protection, 2) pro-environmental behaviors, 3)
perceived local environmental problems, and 4) the NEP scale. Following that, a structural
equation model is developed to test hypotheses based on the literature review. Finally, I discuss

45

how the choice of measure greatly influences the obtained results and make the conclusion with
the direction of future research in China.
3.2 Literature Review
Previous researchers have done a lot of work in terms of social bases of environmental
concern, and most of their work was conducted in industrialized countries. By reviewing
different literature sources between the Western societies and the Chinese societies, I
summarized five hypotheses, which have been tested in different ways in many researchers‟
published work (e.g., Van Liere and Dunlap, 1980; Shen &Saijo, 2008; Fransson & Garling
1999).
Firstly, in what is known as the Age Hypothesis, Van Liere and Dunlap (1980) argued that
younger people are more concerned about environmental deterioration than older generations.
Later, Howell and Laska (1992) also continued to provide support for this hypothesis. One
explanation for the age hypothesis, at least in the USA, is the age-group difference (Malkis and
Grasmick, 1977). It states that compared to older people, younger people are less integrated into
the American economic and cultural systems. Since the solutions to environmental problems
require a radical social reform in traditional economic, social, and cultural systems, it is logical
to expect that younger people can accept such changes more easily.
However, based on a survey of urban residents in Shanghai, China, Shen and Saijo (2008)
stated that age has a positive relation to environmental attitudes, which implies that older
generations are more concerned about the environment than the younger population. The authors
argued that since old Shanghai residents suffered serious environmental problems due to heavy
pollution during 1980s and early 1990s than more recently, they cared more about the
environment than younger generations. Luo and Deng (2008) used the 2000 version of the NEP

46

scale to examine the socio-demographic predictors of visitors to a national forestry park in
China. Their results also indicated that older females and higher educated respondents have more
environmental concern, as reflected by their higher NEP scores. Through surveying urban
residents in China, Xiao and Hong (2010) found that age has a positive association with
environmental concern measured by the NEP scale, when they controlled other variables.
The second hypothesis is the Gender Hypothesis. A number of studies have targeted the
gender difference with regard to environmental attitudes. Based on a series of studies on
industrialized societies, women have been reported to be more concerned about general
environmental issues than men (McStay and Dunlap 1983, Stern, Dietz and Kalof, 1993; Mohai,
1992, Zelezny et al., 2000; Hunter et al. 2004). One widely used theory regarding the
relationship between gender and environmental concern is called gender socialization theory.
This theory states that early childhood socialization leads women to become the role of
caregiver. With this role in their mind, women tend to care about the health and welling-being of
family members and which leads them to a higher level of environmental concern. In contrast,
through early childhood socialized experiences, men, who are assigned the role of the primary
family “breadwinner” are more likely to care about economic issues (Wehrmeyer and McNeil,
2000). However, through examining data collected from the state of Kentucky in the USA,
Arcury and Christianson (1990) found that men were more environmentally concerned than
women. Later, in a cross national analysis, Hayes (2001) argued that although men and women
have differences in environmental knowledge, this has little or no effect on their attitudes toward
environment.
Recent research by Shen and Saijo (2008) however, has shown that women in Shanghai seem
less concerned about the environment than men. Using the 2003 general social survey data, Xiao

47

and Hong (2010) demonstrated that Chinese women expressed lower levels of concern,
measured by the NEP scale, than men when controlling other socio-demographic variables and
environmental knowledge. However, women are also more involved in environmental behaviors
inside of the home, such as recycling, than men. Following the mixed findings about the gender
effect, it makes me ask if there are other mechanisms causing the mixed gender effect under the
Chinese society.
The third hypothesis is the Residence Hypothesis. This hypothesis is that urban residents are
more environmentally conscious than rural residents. Earlier studies on environmental concern
have shown that urban residents exhibited greater concern. Dunlap and Van Liere (1978) and
Fransson and Gä
rling (1999) gave one possible explanation, arguing that this could be because
urban residents are more exposed to the signs of environmental deterioration, such as air and
water pollution, therefore, they have more urgent need to prevent environmental deterioration.
In my view, there are two more explanations for this hypothesis. First, urban residents are
more likely to receive environmental information from mass media than those in rural areas.
Second, rural people are usually economically more dependent on mining and agriculture, and to
perform these activities in a way that can be very harmful to the environment, than are urban
residents. For example, in some rural areas of China, it is still very common for farmers to use
wood to do some cooking.
Other studies have challenged such rural-urban differences in environmental concern. Recent
work by Jones, Fly, and Cordell (1999) indicated that there are no significant differences
between urban and rural residents of Southern Appalachian region on issues related to
environment. As our sample is restricted to urban areas in China, I do not test residence
hypotheses here. Based on available information in the data, I hypothesize that those living in

48

large metropolitan areas are significantly more environmentally concerned than people in the
small or middle size counties or cities. Similar results can be found research in western societies
(Howell and Laska, 1992). However, it still needs to be verified in the Chinese social context.
The fourth hypothesis is the Socio-economic Status Hypothesis. This hypothesis states that
education and income have a positive relationship with environmental concerns. The higher
education and income people have, the more likely they are to be environmentally conscious.
One possible explanation of this is based on Maslow‟s theory hierarchy of needs (Maslow, 1970)
and post-materialist values (Inglehart, 1995). It argues that only when basic material needs, such
as food, shelter, and economic security, are met, can people pursue higher needs such as
environmental quality. It also states that people with "post-materialist" values - emphasizing selfexpression and quality of life - are much more likely to give high priority to protecting the
environment than those with materialist values. Another possible explanation is that highly
educated people are more likely to be in education, research, and high tech jobs that do not
involve environmental destruction. This means that people who don‟t want to destroy the
environment are more likely to go into such jobs and that if you are in such a job, there will be
less economic incentive to destroy the environment..
In terms of effects of occupational prestige on environmental concern, different scholars
draw mixed conclusions from different sources. First, people with high social class occupations
make more money and have more free time. Therefore, those people are expected to have more
concern than low classes. Most studies has found occupational prestige has a positive association
with environmental concern, though such association is light and it is hard to reach a conclusion
that such relationship really existed (Van Liere and Dunlap,1980). For example, Dillman and
Christenson (1972) argued that there are moderate relationships between prestige and

49

environmental concern. Second, as individuals with some occupations might have more chances
to access polluted environment than other occupations, those people might express more concern
than other people in less polluted working condition due to their suffering environmental
pollution. However, such relationships have received limited attention in the empirical research.
It deserves further research in the Chinese context.
Most studies in the West have found a positive relationship between education and
environmental concern (Scott& Willits 1994; Klineberg et al., 1998). However, as for the
relationship between income and environmental concerns, the findings have created controversy.
Arcury (1990) argued that income is significantly correlated to both environmental knowledge
and environmental attitudes. However, based on a 1992 international survey, Dunlap and Mertig
(1995) stated that wealth is negatively related to citizen environmental concern. In China,
different scholars also reached different results based on the different samples and methods. For
example, through using the data collected from Shanghai city, Shen and Saijo (2008) found that
households with higher income have reacted positively to environmental concern, while Xiao
and Hong (2010) used a national urban sample and concluded that income has no influence on
environmental concern.
The final hypothesis is the Political Orientation Hypothesis. It states that liberals have a
more environmental concern than conservatives. Dunlap (1975:432) offered three possible
explanations for this assumption. First, business and industry, which are major supporters of
conservatives, are against environmental reforms due to the costs involved. Second,
environmental reform entails an extension of government activities and regulation, which
conservatives are against. Finally, conservatives do not like innovative actions which are

50

required to launch environmental reform. Later research by Howell and Laska (1992) has also
supported this assumption.
However, in the social context of China, such a political classification about people‟s
political orientation does not exist. Based on China‟s situation, I grouped our respondents as
“Communist Party member” or “non-Chinese Communist Party member”. In my preliminary
analysis, I found there were no significant differences between the environmental attitudes of
Chinese Communist Party members and non Communist Party members.
As a whole, the studies mentioned above focus mainly on those individuals in developed
countries, especially in the United States. China, the largest developing country in this world, has
still received little attention on the relationships between various socio-demographic
characteristics and environmental concern. In 2007, the Central Committee of Communist Party
of China proposed the “Scientific Outlook on Development”, which requires a sustainable
development in China. Research on environmental concern in China has begun to grow, partly
due to that fact that the Chinese government is increasingly emphasizing the importance of
environmental protection. However, compared to a large body of similar research in the West,
we still know little about environmental concern in China. Therefore, using a nationally urban
sample, I have a chance to take this issue into consideration in a more systematic way.
The review above indicates that researchers have had inconsistent results in explaining the
social bases of environmental concern, whether in the West or in China. Although existing
scholarship in China is limited in scope and number, findings from this Chinese based research
are difficult to compare with each other or the studies from other settings due to differences in
measures of environmental concern. For example, Shen and Saijo (2008) classified
environmental concern into three dimensions: concern about general and global environmental

51

problems, concern about local environmental problems, and pro-environmental behavior
measures, while Xiao and Hong used only the NEP scale to represent measurement of
environmental concern.
The mixed findings of the social demographic correlates of environmental concerns whether
in the West or in China, reflects the idea that it is necessary to build comparable and systematic
measures in environmental concern. For example, in terms of substantive issues measured in
environmental concern, researchers can classify diverse issues, such as, air and water pollution,
global warming, and acid rain, into global measures and local measures of environmental
concern. Therefore, to improve researchers‟ ability to explain the social bases of environmental
concern, one point should be kept in the mind: environmental concern cannot be regarded as a
general concern on environmental quality. To examine its social bases, we should focus on each
measures of environmental concern.
3.3 Data and Method
The data for this study was drawn from the 2003 General Social Survey of urban residents
of the People‟s Republic of China. The sample was obtained through a five stage probability
procedure resulting in the selection of 5980 urban residents. The survey includes two parts : Part
A and Part B. The questionnaire consisted of a socio-economic (questionnaire A) and an
environmental (questionnaire B) part.

The number of respondents who answered both

questionnaire A and B is 5073. In this sample, 48.2% are male and 51.8% are female. The
respondents‟ average age is 43.51 years and the average education level is high school (10.44
years).
To examine whether the selection of measure of concern has an effect on the social bases of
environmental concern in China, I proposed a structural equation model here. This model

52

includes the measurement part and structural part. In the measurement part, there are 4 latent
factors of environmental concern and each factor has its own corresponding items and residual
variance (z1-z4). In the structural parts, I add 6 social demographic and economic variables to
each of the four factors (See the Figure 3-1). Thus, those six social demographic variables are
hypothesized to have direct impacts on each of those four factors.
Figure 3-1 Socio-demographic characteristics of Environmental Concern
z1
1

1
1

item1

e1
1

item2

Econ trade
off

e2
1

item3

e3
1

z2

1

1

Age

item4

e4
1

item5

e5
1

item6

Perceived
local problems

Male

e6
1

item7

e7
1

Residence

item8

e8
1

Education

item9

z3
1

Income

e9
1

1

item10

e10
1

item11

Occupation

PEB

e11
1

item12

e12
1

item13

e13
1

z4
1

1

item14

e14
1

item15

e15
1

item16
NEP

e16
1

item17

e17
1

item18

e18

In chapter 2 (paper one), I have constructed a measurement model of environmental concern
based on the same data. To examine the social base of environmental concern, I retrieve the
53

measurement model I built in chapter 2 (paper one). This measurement model includes four
latent construct: tradeoffs between economic growth and environmental protection (ECON
TRADEOFF), pro-environmental behaviors (PEB), the perceived local environmental problems
(PERCEIVED LOCAL PROBLEMS), and the NEP.
Tradeoff between economic growth and environmental protection is one widely used
measure in environmental concern.

This measurement asks respondents to make a choice

between environmental protection and economic security, such as keeping jobs and economic
growth. Klineberg et al. (1998) pointed out that the popularity of this measure is due to its
inclusion in the General Social Survey in the United States. In 2003, this measure was first
introduced to the General Social Survey in China. This measure consists of two indicators. The
first indicator is, “If a certain place faces the dilemma of developing economy and protecting
environment, which of the following view answers for your opinion better?” The percentage
frequencies for the answers to this question are the following: 24.5% of respondents choose
“developing economy comes first”, 2.5% of respondents said they are not sure on this issue, and
73% of respondents prefer “protecting environment comes first”. The second indicator is,
“Suppose in a certain community there is a factory that enjoys economic success and brings
many economic benefits to the local residents, but the factory also produces wastewater that
pollutes the riverhead of downstream areas. If you were one of the residents in the community
where the factory rests, which of the following attitude answers stand for you better? ” There are
also three answers for this question: 1) disagree with closing the factory; 2) not sure; 3) agree to
close the factory. The percentage frequencies for the answers to this question are the following:
13.3% of respondents choose “disagree to close factory”; 1.4% of respondents said they are not
sure on this issue; 85.3% of respondents choose “agree to close factory”.

54

The second type of environmental concern asks respondents if they conducted the following
pro-environmental behaviors. In the survey, it listed 10 questions. Due to some low factor
loading, I kept only 3 out 10 items. This measure includes three indicators: 1, donate money for
environmental protection; 2, actively participate in environmental campaigns sponsored by
government and workplace; 3, actively participate in environmental protection activities held by
Non-governmental organization. (For the exact wording of these items, see Table 3-1)
Table 3-1 Three Items of Pro-environmental Behaviors
Never
(%)

Sometimes Often
(%)
(%)

Donating money for environmental protection
69.1
27.7
Actively participate in environmental campaigns
sponsored by government and workplace
58.7
30.4
Actively participate in environmental protection activities
held by Non-governmental organization
76.1
18.9
Note: Missing data was minimal (less than 0.5%) and recorded as “Sometimes”

3.2
10.9
5

The third type of environmental concern is called the perceived local environmental
problems. A confirmatory analysis shows that 3 items have a low factor loading, so I kept 7 out
of 10 items. It asks respondents if their living communities have the following environmental
problems: air pollution, water pollution, noise pollution, industrial waste pollution, urban
garbage pollution, green space deficiency, and destruction of natural vegetation in forests. The
answers for 7 items are the following: 1, Not serious at all, 2, Not serious, 3, Unsure, 4,
Somewhat serious and 5, Very serious. (For the exact wording, see Table 3-2).

55

Table 3-2 Seven Items of the Perceived Local Environmental Problems
1=Not serious at all; 2=Not very
serious;3=Unsure;4=Somewhat serious,5=Very serious
1(%)
2(%)
3(%)
4(%)
5(%)
3.8
16.8
32.3
28.9
18.1
4.3
18.6
35.9
26.3
14.8
3.9
16.8
35.5
27.2
16.6
6.2
16.9
46.6
18.6
11.6
4.4
17.1
33.1
28.8
16.7
4.3
12.5
43.3
24.9
15.0

1. Air pollution
2. Water pollution
3. Noise pollution
4. Industrial waste pollution
5. Urban garbage pollution
6. Green space deficiency
7. Destruction of natural
vegetation in forests
4.1
10.2
63.9
13.9
7.9
Note: Missing data and those who answer “don‟t know” or “don‟t exist” are recoded as “Unsure”.
A final and common way of measuring environmental concern is to use the NEP scale. It
asks relatively abstract questions to explore the interrelationships between human activities and
ecosystems. The original NEP scale had 12 items. Later, Dunlap and Van Liere (2000) revised it
to include 15 items. However, a confirmatory factor analysis of the 15 items showed that 9 items
had low factor loadings. Therefore, I kept 6 items out of 15 items, which also can help us make
comparisons with similar studies in the U.S (Table3- 3)
Table 3-3 Six Items of New Environmental Paradigm Scale Items and Percentage
Frequencies in China and U.S
1=Strongly disagree; 2=Mildly disagree
3=Unsure; 4=Mildly agree; 5=Strongly
agree
NEP Items
1.When humans interfere with nature it often
produces disastrous consequences
2.Humans are severely abusing the environment
3. Despite our special abilities humans are still
subject to the law of nature
4. The earth is like a spaceship with very limited
room and resources
56

%
China
US
China
US
China
US
China
US

1
2.7
2.5
3.6
1.5
1.1
0.8
1.7
4.8

2
4.6
11.2
8.9
9.3
5.1
2.9
6.5
13.4

3
11.1
4
13.2
2.6
19.3
5.4
26.4
7.5

4
31.3
37.6
33.9
35.3
39.5
31.3
33.1
36.3

5
50.4
44.6
40.4
51.3
35
59.6
32.3
38

Table 3-3 (cont’d) Six Items of New Environmental Paradigm Scale Items and Percentage
Frequencies in China and U.S
NEP items
%
1
2
3
4
5
25.1 36.9 29.7
5. The balance of nature is very delicate and easily China 1.7 6.5
upset
US
1.4 14.1 5.9
32.8 45.9
18.7 32.9 39.1
6. If things continues on their present course, we China 2.2 7.1
will soon experience a major ecological catastrophe US
3.6 14.1 16.9 31
34.3
Notes: data from China are recoded so that higher scores show more support for the NEP. Data
from U.S. is from the paper by Dunlap et al.(2000)
Do Chinese respondents appear to show more environmental concern than in the U.S and
other developed countries? From Table 3-3 above, it seems that the U.S respondents have higher
percentages agreeing with 4 out of 6 NEP items than their Chinese counterparts.
What if using the NEP measured by 15 items? According to CGSS2003 data, the total score
in the NEP scale is 51.7 out of 75; while using 1995 survey data on environmentalism, Gunnar
(1999) found that the total score in the NEP is 55.36 out of 75 in the U.S. Huddart-Kennedy et
al. (2009) used data from a national survey in Canada and found that the average score in the
NEP from both rural and urban areas is 55.05 out of 75. Considering CGSS2003 excluded
respondents in rural areas, it can be inferred that Chinese respondents might have a lower
environmental concern than US and many developed countries.
Our socio-demographic variables include gender (female=0 vs. male=1), age (continuous
variable ranging from 18 to 72), and individual socio-economic variables consist of education
(years of schooling-- those who did not get any school education are coded as 0; those who only
finished primary school are coded as 6; those who finished middle school are coded as 9; those
who have high school or middle professional school diploma are coded as 12, those who have
associate degree are coded as 15; those who have a bachelor degree are coded as 16 year and
those who have a post graduate degree or above are coded as 19 years),
57

Other variables are income (continuous variable), The average income of urban residents in
China is 10,000 yuan (1 U.S.D=8.2 yuan in 2003), and occupation status (0，those without any
occupation, 1, those who neither have management positions nor professional titles as nonmanagement and non professional and 2, those who either have management position or
professional technique titles). Finally, our residency variable is a categorical variable (town
level=0, county level=1, city level=2, capital city level=3) (See Table 3-4 for the details).
Table 3-4 Social Demographic and Economic Variables
Variables
Age
Gender
Community Size
Education
Income of a year (Yuan)
Occupation status

Description of Variables
Mean
Continuous Variable (year)
43.51
Female=0, Male=1
0.48
Small town=0, County level city=1,
Median size city=2, Large city=3
2.03
Continuous Variable （year）
10.44
Continuous Variable (thousand unit)
10.01
No occupation=0
Non-management and non professional
occupation=1,
Management and professional occupation=2 0.67

S.D
13.17
0.49
1.03
3.69
12.83

0.71

3.4 Results
The results of the structural equation model (Table 3-5) indicate that social, demographic,
and economic variables have different effects on each of these four factors. In terms of
explanatory power, these variables together explain no more than 12.6% of variance in any of the
four factors, which is similar to the findings from the Western setting (Marquart-Pyatt, 2008). It
should be noted here that these four factors are not highly correlated. Among all four factors,
social, demographic, and economic variables account for the highest percentages of variance in
the perceived global problems factor ( R 2 =0.12), while they explain the least variance in the

58

economic tradeoff factor ( R 2 =0.031). In terms of goodness-of-fit indices, Chi-square value is
1807.85 with 219 degrees of freedom and p-level less than 0.001. CFI and RMSEA are 0.927
and 0.038, respectively which implies good model fit.
Table 3-5 Standardized Regression Coefficients on Each of Four Factors
Economic
Trade off

Perceived
Local problems

Age
0.092***
Male
0.014
Community size
-0.038
Education
0.178***
Income
0.021
Occupation
-0.005
2
0.031
R
Note ：*p<=0.05, **p<=0.01, ***p<=0.001

0.011
0.035*
0.129***
0.024
-0.015
-0.022
0.02

PEB

NEP

-0.011
-0.026
0.062***
0.23***
0.082***
0.109***
0.089

0.076***
0.095***
0.023
0.338***
0.018
0.002
0.126

The second column of Table 3-5 shows that age has a positive relationship to economic tradeoff factor and education. This means that older people are more likely to prefer
environmental protection to economic growth than younger people. Education has the strongest
effect on the factor of economic tradeoff (b=0.178). It is interesting to find that the size of the
residential community was not significantly associated with the economic tradeoff factor. It
might reflect the fact that ordinary urban residents, whether living in big city or small city,
cannot do much in environmental protection in terms of public policy decision making. As for
the other variables, such as gender, income, and occupation status, there is no significant
association with the economic tradeoff factor.
Contrary to the result in the second column, the third column indicates that the size of the
residential community is significantly positive related to the factor of perceived local problems.
People living in the big cities perceive more severe environmental problems due to two reasons.
First reason is that the industrial development strategy planned by the Chinese government after
59

the foundation of People‟s Republic of China followed the road of the former Soviet Union and
then large cities usually had priority to develop heavy industry. Although this situation has been
changed in recent years, the consequences of heavy industry in large cities are still evident. For
example, air pollution due to steel factories built inside the city. Second is that due to mass
media‟s fast development in big cities, residents of larger cities have easy access to
environmental information. In addition, it has the strongest effect on the perceived local problem
factor relative to other variables. This finding is essentially the same as Tremblay and Dunlap
(1978) conclusion that residential differences in concern are most pronounced when the
environmental issue studied was of local concern.
Age and gender have no significant association with pro environmental behaviors (PEB)
factor. In terms of the gender effect, the result suggests that males seem to perceive more local
environmental problems than females. The reason might be related to different social roles in the
family. It was argued that women took greater share of domestic duties, which might reduce their
availability of perception of different kinds of environmental problems, while men used to go out
for work and increase their access to various environmental problems. However, such effect also
does not exist in the PEB factor. Surprisingly, education had no significant association with
perceived local problems, contrary to common findings in the West and Chinese settings. For
example, Klineberg et al. (1998) regressed on different kinds of local pollution and found that
education is a significant predictor of perceived local problems.

In addition, income and

occupation status also have no significant associations with perceived local problems factor.
The fourth column exhibits that all socio-economic variables which include education,
income, and occupation status have a significant relationship to the PEB factor. However, age

60

and gender has no effect on the pro-environmental behaviors. The size of the residential
community has a significant and positive relationship with the PEB factor.
Why does community size have a positive association with the perceived local problems
factors? One possible explanation for it is that there are more urgent environmental problems
objectively existing in their lives, while those individuals in the large cities subjectively perceive
more environmental problems due to easily accessing environmental information through mass
media. Such patterns also exist in the perceived local problem factor. Regarding gender effect,
there are no statistically significant differences in the PEB factor. The socio-economic status
variables: income, education, and occupation ranking, have a positively significant association
with the PEB factor, and like the effect of economic -trade off factor, education becomes the
strongest predictor again.
The fifth column demonstrates the social bases of the NEP factor. Different from the
finding in the PEB factor, age is found to be positively related to the NEP factor. The analysis
suggests that an older individual has a higher NEP factor score than a younger person. As for
gender effect, males seem to have a higher score on the NEP factor than females. These two
results are similar to the findings provided by Xiao and Hong (2010). Different than its role in
other factors, the size of residential community does not play any significant role on the NEP
factor and economic trade off factor. Like most findings in the West, income and occupation
status has no effect on the NEP factor. Compared to other variables, education is still the
strongest predictor in the NEP factor.
3.5 Discussion
Applying the national data collected in urban China, I first constructed a measurement
model of environmental concern and then examined how the selection of particular measure of

61

environmental concern influences social demographic characteristics through the structural
equation model. There are many reasons for inconsistent findings on the social bases of
environmental concern, such as methodological issues, different measurement models, and
analytical tools. Here, the results revealed that the selection of particular measure of
environmental concern does affect its relation to social-demographic variables.
In terms of age hypothesis, our finding is different from the findings by (Howell and Laska,
1992, Van Liere and Dunlap, 1980), but it is consistent with the conclusion by Shen and Saijo
(2008).

Gender is consistently correlated with environmental concern across the NEP and

perceived local problems factors. In terms of gender effect, the result here suggests that males
have more environmental concern than females in urban China, which is different from some
previous research results in Western societies. (Schahan and Holzer, 1990; Mohai, 1992; Hunter
et al., 2004). However, this result fits with Marquart-Pyatt (2008)‟s research in the former
communist countries and Shen and Saijo (2008) conclusion based on the data collected from
Shanghai in China. Does it mean that some source of environmental concern is related to the
countries‟ ideological differences and developing levels? This question needs further research.
Regarding education hypothesis, the result exhibits that the more educated person generally has
more concern for the environment, which is parallel with most previous research (Dunlap et al.,
2000; Jones and Dunlap, 1992, Klineberg et al.,, 1998; Weaver, 2002).
For some demographic variables, the data presented here tell different stories. Age is
positively associated with the econ tradeoff factor and the NEP factor, while has no relation to
the factor of the PEB. This suggests that some aspects of environmental concern might be more
apparent in some specific age groups while less obvious in other age groups. The type of
residential community has a significant association with the PEB factor and perceived local

62

problems factor among four selected measures. The result here shows that individuals in the
larger cities are slightly more likely to conduct pro-environmental behaviors at the same time.
Individuals in the large cities also perceived more local environmental problems. Different
effects of age and residence on the selected measures also suggest that the selection of certain
measure of environmental concern really matters (Dunlap and Jones, 2002). Through crossnational studies, Marquart-Pyatt (2008) also reinforced this argument.
Except for their effects on the PEB factor, income, and occupation status have little effect on
the other measures of environmental concern. The significant associations between income,
occupation status and the PEB factor may reflect the fact that environmental quality is usually
treated as a luxury good, and the privilege of upper and middle classes (Van Liere and Dunlap,
1980). The result here illustrates that people with high income and occupation status are more
likely to conduct pro-environmental behaviors. However, it is surprising to note that education
has no association with the perceived local problems factor. This might reflect that perceiving
local problems is essentially a process of what they see and experience by themselves. This
process is not a matter of individual‟s education level. Due to data limitation, I did not construct
perceived global problems factor as some researchers did. Is individual‟s education level closely
related to perceived global problem because only literacy individual can easily access the mass
media such as newspapers, TVs and Internet? This question warrants future research.
Theorizing measures and models, which capture individual level and context-level sources of
environmental concern, may be essential for future research. Inglehart (1995) argued that sources
of concern for the environment are the combination of “objective problems” and “subjective
values”, he attributed concern for the environment as resulting from objective problems
(environmental degradation). These context-level sources are also considered as objective reality.

63

Through the survey data collected from Israel‟s urban centers, Drori and Yuchtman-Yaar (2002)
concluded that urban public‟s concern for the environment is responsive to the environmental
vulnerability of its community. For example, due to different industrial structures and
environmental regulations, air pollution might differ across contexts. Therefore, to further
understand the social bases of environmental concern, some context- level variables might be
considered when examining the sources of environmental concern. This kind of analysis requires
more advanced statistical tools.
3.6 Conclusion and Summary
In sum, this research is not only systematically study of social determinants of
environmental concern measured by four dimensions in urban China, but also examines some
significant differences between urban Chinese residents and their western counterparts. The
results suggest that more attention should be paid to the selection of particular measure of
environmental concern. The data used here clearly demonstrates that it really makes a difference
how environmental concern is measured (Van Liere and Dunlap, 1981; Klineberg et al., 1998;
Samdahl and Robertson, 1989). This study thus provides a baseline for future research to explore
possible driving forces of environmental concern in urban China.
This study also has some obvious limitations. I would like to suggest several areas for future
research in China. First, four dimensions are used here to measure environmental concern in
urban China. There are still other ways to conceptualize environmental concern. For example,
Morrissey and Manning (2000) used environmental values and ethics as an alternative measure
of environmental concern. Future studies employing greater numbers of dimensions for
environmental concern will be helpful to ensure the validity of the findings. Second, as economic
and social transformations are still under way in modern China, it will be useful to examine

64

whether the meaning of environmental concern is also changing over time. Since this study is
cross sectional, it cannot verify this possibility. Future research should consider this point.
Finally, rural and urban places may exert different influences on environmental concern (Arcury
and Christenson 1993; Fortmann and Kusel 1990, Van Liere and Dunlap 1980). This urban
sample makes it impossible to explore urban and rural differences of environmental concern in
China. It will be helpful for continued research to collect samples from both rural and urban
areas in China.

65

BIBLIOGRAPHY

66

BIBLIOGRAPHY

Albrecht, D., Bultena, G., Hoiberg, E., & Nowak, P. (1982). The New Environmental Paradigm
Scale. Journal of Environmental Education, 13, 39-43.
Arcury, T. A. (1990). Environmental Attitude and Environmental Knowledge. Human
Organization, 49, 300-304.
Arcury, T. A., & H.Christianson, E. (1990). Environmental Worldview in Response to
Environmental Problems: Kentucky 1984 and 1988 Compared. Environment and Behavior,
22(3), 387-407.
Buttel, F. H., & Flinn, W. L. (1978). Social Class and Mass Environmental Beliefs: A
Reconsideration. Environment and Behavior, 10, 433-450.
Buttel, F. H., & W.L.Flinn. (1976). Economic Growth Versus the environment: Survey
Evidence. Social Science Quarterly, 57, 410-420.
Drori, I., & Yuchtman-Yaat, E. (2002). Environmental Vulnerability in Public Perceptions and
Attitudes: The Case of Israel's Urban Centers. Social Science Quarterly, 83(1), 53-63.
Dunlap, R. E. (1975). The impact of political orientation on environmental attitudes and actions.
Environment and Behavior, 7, 428-454.
Dunlap, R. E., & G.Mertig, A. (1995). Global Concern for the Environment: Is Affluence a
Prerequisite? Journal of Social Issues, 51(122-137).
Dunlap, R. E., & Jones, R. E. (2002). Environmental Concern: Conceptual and Measurement
Issues. In R. E. Dunlap & W. Michelson (Eds.), Handbook of Environmental Sociology (pp. 482524). Westport, CT: Greenwood Press.
Dunlap, R. E., & Valn Liere, K. D. (1978). The "new environmental paradigm": A Proposed
Measuring Instrument and Preliminary results. Journal of Environmental Education, 9, 10-19.
Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). Measuring Endorsement
of the New Ecological Paradigm: A Revised NEP Scale. Journal of Social Issues, 56(3), 425442.
Fransson, N., & Gä
rling , T. (1999). Environmental Concern: Conceptual Definitions,
Measurement Methods and Research Findings. Journal of Environmental Psychology, 19, 369382.
Hayes, B. C. (2001). Gender, Scientific Knowledge, and Attitudes toward the Environment: A
Cross-National Analysis. Political Research Quarterly, 54(3), 657-671.
Hong, D. (2006). Measurement of Environmental Concern: NEP Scale's Application in China.
67

Chinese Journal of Society, 5, 71-91.
Howell, S. E., & Laska, S. B. (1992). The changing face of the environmental coalition: A
Research Note. Environment and Behavior, 24(134-144).
Hunter, L. M., Hatch, A., & Johnson, A. (2004). Cross-National Gender Variation in
Environmental Behaviors. Social Science Quarterly, 85(3), 677-694.
Inglehart, R. (1995). Public Support for Environmental Protection: Objective Problems and
Subjective Values in 43 Societies. PS: Political Science and Politics, 15, 57-72.
Jones, R. E., Fly, J., & K.Cordell. (1999). How green is my valley? Tracking rural and urban
environmentalism in the southern Appalachian Ecosystem. Rural Sociology, 64(3), 484-499.
Jones, Robert Emmet and Riley E.Dunlap. 1992. "The social Bases of Environmental Concern:
Have They changes Over Time?" Rural Sociology 57:28-47.
Klineberg, S. L., McKeever, M., & Rothenbach, B. (1998). Demographic Predictors of
Environmental Concern: It Does Make a Difference How it's Measured. Social Science
Quarterly, 79(734-753).
Lounsbury, J. W., & Tornatzky, l. G. (1977). A Scale for assessing Attitudes toward
Environmental quality. Journal of Social Psychology, 101, 299-305.
Luo, Y., & Deng, J. (2008). The new environmental paradigm and nature-based tourism
motivation. Journal of Travel Research, 46, 392-402.
Malkis, A., & Grasmick, H. G. (1977). Support for the Ideology of the Environmental
Moverment: Tests of Alternative Hypotheses. Western Sociological Rview, 8, 25-47.
Maloney, M. P., & Braucht, G. N. (1975). A Revised Scale for the Measurement of Ecological
Attitudes and Knowledge. American Psychologist, 30, 787-790.
Marquart-Pyatt, S. T. (2008). Are There Similar sources of environmental concern? Comparing
Industrialized countries. Social Science Quarterly, 89(5), 1312-1335.
Maslow, A. H. (1970). Motivation and personality. New York: Harper Press.
Mobley, Catherine, Wade M. Vagias, and Sarah L. DeWard. (2010). "Exploring Additional
Determinants of Environmentally Responsible Behavior: The Influence of Environmental
Literature and Environmental Attitudes." Environment and Behavior 42:420-447.
McStay, J., & Dunlap, R. E. (1983). Male-Female Differences in Concern for Environmental
Quality. International Journal of Women's Studies, 6(291-301).
Mohai, P. (1992). Men, women and the environment: An examination of the gender gap in
68

environmental concern activism. Society and Natural Resources, 5, 1-19.
Mohai, Paul, Solange Simoes, and Steven R. Brechin. (2010). "Environmental Concerns, Values
and Meanings in the Beijing and Detroit Metropolitan Areas." International Sociolgy 25:778817.
Morrissey, J., & Manning, R. (2000). Race, Residence and Environmental Concern: New
Englanders and the White Mountain National Forest. Human Ecology Review, 7(1), 12-23.
Samdahl, D. M., & Robertson, R. (1989). Social derminants of environmental concern:
Specification and test of the model. Environment and Behavior, 21, 57-81.
Schahn, J., & Holzer, E. (1990). Studies of individual environmental concern: The role of
knowledge, gender and background variables. Environment and Behavior, 22, 767-786.
Scott, D., & Willits, F. K. (1994). Environmental Attitudes and Behaviors: A Pennsylvania
Survey. Environment and Behavior, 26(2), 239-260.
Shen, J., & Saijo, T. (2008). Reexaming the relations between socio-demographic characteristics
and individual environmental concern: Evidence from Shanghai data. Journal of Environmental
Psychology, 28, 42-50.
Stern, P. C., Dietz, T., & Kalof, L. (1993). Value orientations, gender, and environmental
concern. Environment and Behavior(25), 322–348.
Tognacci, L. N., Weigel, R. H., M.F.Wideen, & Vernon, D. T. A. (1972). Environmental
Quality: How Universal is Public Concern? Environment and Behavior, 4, 73-86.
Tremblay, K. R., & R.E.Dunlap. (1978). Rural-urban residence and concern with environmental
quality: A replication and extension. Rural Sociology, 43(3), 474-491.
Van Liere, K. D., & Dunlap, R. E. (1980). The Social Bases of Environmental Concern: A
Review of Hypotheses, Explanations and Empirical Evidence. Public Opinion Quarterly, 44(2),
181-197.
Van Liere, K. D., & Dunlap, R. E. (1981). Environmental Concern: Does It Make a Difference
How It's Measured. Environment and Behavior, 13, 651-676.
Weaver, A. A. (2002). Determinants of environmental attitudes: A five county comparison.
International Journal of Sociology, 32, 77-108.
Wehrmeyer, W., & McNeil, M. (2000). Activists, pragmatists, technophiles and tree-huggers?
Gender differences in employees' environmental attitudes. Journal of Business Ethics, 28, 211222.
Weigel, R. H., & J.Weigel. (1978). Environmental Concern: the Development of a Measure.
69

Environment and Behavior, 10, 3-15.
Xiao, C., & Hong, D. (2010). Gender Differences in Environmental Behaviors in China.
Population and Environment, 32(1), 88-104.
Xiao, C., & M.McCright, A. (2007). Environmental Concern and Sociodemographic Variables:
A Study of Statistical Model. The Journal of Environmental Education, 38(2), 3-12.
Zelezny, L. C., Chua, P.-P., & Aldrich, C. (2000). Elaborating on Gender Differences in
Environmentalism. Journal of Social Issues, 56, 443-457.

70

Chapter 4
Multi-level Analysis of Urban Chinese Residents’ Environmental
Attitude and Behaviors
4.1 Introduction
For several decades, the Chinese economy has experienced unprecedented growth with an
average annual GDP growth rate of 8%. This rapid economic growth, along with a huge
population in China, has resulted in serious environmental problems, ranging from air and water
pollution to loss of biodiversity and deforestation. For example, about 40% of China‟s land is
affected by soil erosion and deforestation (Wang, 2004). All of these problems not only cause
substantial economic losses, but also threaten the health of Chinese residents. It was estimated
that annual loss from air pollution was 40 billion US dollars (Duan et al., 1993). At the same
time, according to a recent published research, after adjustment for individual level
socioeconomic variables, air pollution increased the odds of developing ADLs (here, activities of
daily living was used to describe difficult level of functioning in performing daily tasks),
cognitive impairment and other health deficits (Zeng et al., 2010).
The reasons behind most of environmental problems mentioned above are closely related to
the behaviors and underlying attitudes of the Chinese people. However, we know little of the
environmental concern and behaviors of the Chinese public, which are important contributors to
environmental impact (Xiao and Hong, 2010). Therefore, studying the underlying forces of
environmental-related behaviors at different levels in China is extremely important for policy
makers and researchers seeking solutions to environmental problems that require behavioral
change.
The current literature has largely focused on the individual level determinants of
environmental attitude and pro-environmental behavior (PEB) from different disciplines or
71

subjects. For example, economists try to examine the influence of external conditions, such as
income, price, and other economic characteristics, on individuals‟ behaviors (Clark et al., 2003).
This approach assumes that individual decisions are the consequence of rational choice.
Sociologists also tend to emphasize the effect of individual‟s social characteristics on attitude
and behaviors. For example, some researchers examine the role of gender, education, political
orientation and residency on pro-environmental behaviors (Zelezny, et al., 2000; Hunter et al.,
2004; Xiao and Hong, 2010). On the other hand, some social scientists focus on the internal
forces of pro-environmental behaviors, such as intention, value, belief, and personal norms.
There are two main theories according to this line of research.
The first main theory is called the Theory of Planned Behavior (TPB), which has developed
from an earlier version of Fishbein and Ajzen‟s (1975) Theory of Reasoned Action. The second
main theory is called Value-Belief-Norm (VNB) theory, which was developed to elucidate the
link between attitude and behaviors (Stern 2000; Stern, Dietz and Guagnano, 1995; Schultz and
Oskamp, 1996). Despite the plethora of studies on environmental attitude and PEB, most of
these studies in the Western societies have some major limitations. First, previous work either
using TPB or VBN mostly focused on the developed countries; published analyses of data from
developing countries, including China, are rare. Second, most previous research collected data
are either from a few areas in one country or from selected developed countries. Third, most of
their work, especially for those applying these two main theories in environmental behaviors,
focused on the individual level determinants of PEB, since these two theories emphasize on
important roles of intention, attitude and value on the formation of individual behavior.
Several scholars have suggested the significant role of contextual level variables such as
local pollution levels and extent of economic development influence environmental attitude and

72

pro-environmental behaviors in the community. For example, Sandar (2008:1331) argued that
“we should also address perceived seriousness of objective conditions as influences on
environmental concerns in formulating regionally-based hypothesis and further explore
contextual explanation.” Through discussing changes in public transport use when individual
relocated to other place, Bamberg (2006) argued that the change in the environment and its
interaction with intervention policies (a free public transport ticket) were responsible for the
changes of behaviors. Through a multilevel analysis of 50 nations, Gelissen (2007) argued that it
is necessary to examine the combined effects of individual level and contextual level factors on
pro-environmental attitudes.
Our main goal in this paper is to identify the effects of individual-level as well as contextual
-level determinants on environmental attitude and PEB in China, through using a national sample
with a nested data structure, which is rare in current environmental social science research. The
following section provides an interdisciplinary literature review on pro-environmental behaviors.
Following that, I offer a detailed description of our analytical models, data and variables. After
that, I present our analyze results. Finally, I conclude with a discussion of insights gained.
4.2 Literature Review
Since the 1970s, there has been an increasing interest among social scientists in the study of
attitudes and behaviors with environmental consequences. These studies not only have
theoretical value, but also have practical value, because policy makers can take actions from
these studies to increase the level of environmental concern and pro-environmental behavior and
thus prevent environmental degradation. Previous research has demonstrated that individuals
with high environmental concern are more likely to conduct pro-environmental behavior. Proenvironmental

behavior

is

also

referred

73

to

as

environmentally-friendly

behavior,

environmentally-responsible behavior and conservation behavior. Cottrell (2003:356) defined
environmentally-friendly behavior as “any individual or group action aimed to do what is right
to help protect the environment in a general daily practice.”There are several types of
environmentally-friendly behaviors. Recent developments on environmentally-significant
behaviors have provided theoretical support for this classification (Tindall et al., 2003; Stern
2000). So it is necessary to understand environmental behaviors in the specific level. In this
research, I will adopt Stern‟s (2000) classification of pro-environmental behavior: private and
public environmental behaviors. Private environmental behaviors represent that those behaviors
take place in the private sphere: “the purchase, use and disposal of personal and household
products that have environmental impact” (Stern, 2000:409). Public environmental behaviors are
those behaviors that happen in the public sphere and attempt to influence public policy, such as
donating money for environmental activities and actively joining environmental organization.
4.2.1 Individual-Level Variables: Socio-demographic Characteristics
Previous research has revealed that age, income, education, and gender are slight or
moderate correlates of environmental attitude and EFB (Dunlap and Van Liere, 1984; Ostman
and Parker, 1987; Hines et al., 1987; Scott and Willits, 1994; Mobley et al., 2010). However,
like the social bases of environmental concern measured by the NEP scale, some effects of
individual characteristics on general or specific environmental behaviors are inconsistent.
In Western settings, most research has shown that age has a significant association with
environmental concern, though environmental concern can be measured by different dimensions.
Theodor and Luloff‟s (2002) research suggested that age is not a significant predictor of proenvironmental behaviors, which include private and public environmental behaviors. Cottrell‟s
(2003) research continued to confirm this argument. Through surveying recreational boaters in
74

Maryland in 1992, Cottrell found that age is not significantly related to general responsible
environmental behavior, though it has a significant effect on environmental concern. However,
based on a national sample in China, Xiao and Hong (2010) provided different results about the
age effect in two types of environmental behaviors. They found that age has a positive significant
effect on private environmental behavior but a negative significant effect on public
environmental behavior.
As for the income effect on environmentally-friendly behaviors, Hines et al. (1987) found
that income has a positive correlation with responsible environmental behaviors in a metaanalysis. Later, through surveying Pennsylvania residents, Scott and Willits (1994)‟s research
continued to support the positive effect of income on pro-environmental behaviors. However,
some research also challenged this argument. For example, Cottrell (2003)‟s result suggested that
income has no significant effect on responsible environmental behaviors. Kasapoğlu and Ecevit
(2003) also found that there are no significant association between income and responsible
environmental behaviors. In terms of public environmental behaviors, Theodori and Luloff
(2002)‟s research indicated that respondents with higher incomes were significantly more likely
than those with lower incomes to contribute money and time to environmental organizations.
However, in China, researchers (Xiao and Hong, 2010) found that individual with a higher
income are more likely to conduct private environmental behaviors (e.g., recycling).
Among social demographic variables, education is reported to have the most consistent effect
on environmental concern and different types of environmental behaviors (Ostman and Parker,
1987; Scott and Willits, 1994; Xiao and McCright, 2007). Most of the research in developed and
developing countries has indicated that individuals with higher levels of education have more
environmental concern and are more likely to engage in environmentally-responsible behaviors.

75

For example, using data collected in four communities in Pennsylvania, Theodori and Luloff
(2002) found that more highly educated respondents were significantly more likely than those
with lower education to conduct both public and private environmental behaviors. In China,
through path analysis, Xiao and Hong (2010) concluded that except for gender, education was
the only predictor with considerable indirect regression weights on environmental behaviors and
environmental concern.
In the past several decades, there has been a huge amount of research on environmental
attitudes or behaviors in the U.S. This includes an analysis of the effect of gender in individual‟s
attitudes and behaviors (e.g., Davidson and Freudenburg, 1996; Hunter et al., 2004; Hayes, 2001;
Dietz et al., 2002). Gender differences in environmental behaviors have aroused more debates
than other social determinants of environmental behaviors. Many studies have indicated that
women tend to have greater levels of participation than men in various environmental behaviors
(Zelezny et al., 2000; Sherkat & Ellison, 2007). In addition, the effects of gender are consistently
stronger on environmental behaviors than on environmental attitudes (Zelezny et al., 2000).
However, other research challenged this argument and stated that there are no significant
gender differences in environmental behaviors (Blankenau et al., 2008; Bergenguer et al., 2003).
Additionally, according to classification of public and private environmental behaviors, some
research in China and U.S. suggested that women are more likely to engage in private
environmental behaviors (recycling, buying/eating organic) as compared to men (Hunter et al.,
2004; Xiao and Hong, 2010), while men are more actively engaged in public environmental
behaviors than women (Blocker and Eckberg, 1997; Davidson and Freudenburg, 1996).

76

4.2.2 Individual Level Variables: Socio-psychological Characteristics
Both TVB and VBN theory emphasize the role of environmental attitudes in predicting
environmental behaviors. The New Ecological Paradigm (NEP) developed by Dunlap and his
colleagues has been widely used to explore the changing attitudes about environment, not only in
the developed countries such as the U.S., but also in developing countries, including China
(Dunlap, Van Liere, Mertig, and Jones, 2000).

Past research has indicated that general

environmental attitude measured by NEP has a mixed effect on environmental behaviors.
According to a meta-analysis by Hines et al. (1987), psycho-social variables, including general
environmental attitude, have a more significant effect than social-demographic variables on
environmental behaviors. However, Scott and Willits (1994) found that the influence of the NEP
on environmental behaviors is very weak or absent. Using the NEP as a measure of general
environmental attitude, the current study aims to provide further clarity to this debate.
Despite the NEP‟s widespread use, some researchers (Frantz and Meyer, 2004; Stern, Dietz
and Guagnao, 1995) have pointed out that the NEP has limitations, due to its focusing on the
general and abstract relationship between humans and nature. Thus, the measure of general
environmental attitude would be better complemented with a measure of specific environmental
concern (Mobley, Vagias and DeWard, 2010). Related works assume that individual behaviors
toward environment should have something to do with what they think and feel with respect to
the environmental problems around them. In this research, perceived seriousness of
environmental problems in local communities is used to measure what people think and feel with
respect to environmental problems around them.

77

4.2.3 Contextual -Level Variables
Despite the wide range of studies investigating individual differences of environmental
attitude and behaviors, only a few of them examine the effect of different contexts such as local
economic development level and local pollution level, on environmental attitude and behaviors.
One of the few examples that do so, Dolnicar and Grü (2009) argued that it is necessary to
n
focus on both inter-individual heterogeneity and context/environmental heterogeneity. Through a
multilevel analysis of 50 nations, Gelissen (2007) concluded that contextual-level variables, such
as GDP per capita, GDP growth rate, are related directly to levels of support for environmental
protection among nations. Based on the past literature in Western societies, I summarize below,
two hypotheses which are related to our contextual level variables. Although the contextuallevel variables in this research only includes macro economic development and industrial
pollution variables, some western scholars (Haller and Halder, 2008) argued that role of the state
and religion, as the contextual level variables, also needs to be considered in the multi-level
analysis. It should be noted that due to data limitation, only two main hypotheses are tested in
this research.
The first hypothesis is called the objective problems hypothesis. This hypothesis comes from
Inglehart (1995)‟s argument that the growth of environmentalism is the combination of
“objective problems” and “subjective values”. Inglehart (1995) also found evidence for the
hypothesis that the public in some developing countries has a high environmental concern
partially due to the existing serious environmental problems. But, this thesis arouses much
debate. Through comparing results from three waves of the “World Values Survey” to those of
the Health of Planet Survey, Dunlap and York (2008) argued that citizen concern for the
environment is neither dependent on national affluence, nor on postmaterialist values. The

78

“objective problems” here referred to various environmental problems such as air pollution and
water pollution. Based on the challenge-response model, this argument stated that people are
more likely to conduct environmentally friendly behaviors because they are directly confronted
with serious environmental problems (Gelissen, 2007).
The second hypothesis is called the affluence hypothesis. From the view of environmental
economics, environmental quality is regarded as a luxury good. It is logical then to expect that
people in wealthier countries have high environmental concern which will lead them to convert
their concern to participate more environmentally friendly behaviors. Several studies found
positive evidence for this hypothesis in the environmental concern and pro-environmental
behaviors research (Diekmann and Franzen, 1999; Franzen, 2003; Glissen, 2007; Franzen and
Meyer, 2009). In a multilevel analysis of 50 nations, Glissen (2007) found that the public in
comparatively wealthy countries has a higher support for environmental protection. Later,
through the ISSP (International Social Survey Program) data, Franzen and Meyer (2009) have
indicated that more concern is reported in wealthier countries than in poor nations. Through
analyzing the 2000 ISSP data, Freymeyer and Johnson (2010) found that respondents living in
wealthier countries, no matter the environmental quality, more frequently join public proenvironmental behaviors such as joining environmental groups, signing petitions concerning the
environment and giving money for environmental causes.
These two hypotheses have aroused some debate in the studies of environmental concern. For
example, through investigating the data from the Health of Planet Survey, Dunlap and Mertig
(1995) argued that environmental concern has no significant association with a country‟s wealth.
Despite some debate about these two hypotheses, four conclusions can be drawn from the
literature related to these two hypotheses. First, even if these two hypotheses could find evidence

79

in the study of environmental concern, further systematic investigations are needed to see where
these two hypotheses apply. Second, although environmental concern or attitude can be one of
the driving sources of environmental behaviors, we still know little about how much the degree
of environmental concern or attitudes can be translated into an action. Third, most of the current
multilevel analysis for these two hypotheses focuses on two levels: the individual level and the
cross-national level. However, it is well-known that each nation or country has its own cultural,
historical, and economic background. Using data collected from the different cities in the single
nation setting can help us to rule out the possibilities which might happen in the analyses of data
collected from different countries. That is, other factors, such as different cultural, historical and
economic variables might deviate the research results. Fourth, few of the studies that have
examined the role of contextual level variables, have also proposed an analytical approach that
integrated both the individual and contextual bases of environmental attitudes and proenvironmental behavior in developing countries, especially in China.
All in all, to examine possible multilevel effects involving individual and contextual-level
variables on environmental attitudes and pro-environmental behaviors, I test several sets of
hypotheses. Individual level hypotheses variables include the effects of individual level variables
(age, income, gender, and education), one general environmental attitude and perception of local
environmental problems. Contextual-level hypothesis include affluence and “objective problem”
hypothesis.
Therefore, the detailed hypotheses in individual and contextual level are following:
Individual level hypotheses (all with effects of other variables being controlled): women have
more pro-environmental attitudes than men. Women are more likely to conduct private
environmental behavior while men are more likely to conduct public environmental behavior.

80

Personal income, age, education attainment, NEP and perception of local environmental
problems are positively related to pro-environmental attitudes and public and private
environmental behaviors. Contextual level hypotheses (all with effects of other variables being
controlled): city type, GDP per capita, and percent GDP from the primary sector are positively
related to environmental attitudes and public and private environmental behaviors (the affluence
hypotheses). Industrial waste water, industrial sulfur dioxide, and industrial dust are positively
related to environmental attitudes and public and private environmental behaviors respectively
(objective problems hypotheses).
4.3 Data Sources
The first dataset I use comes from the Chinese General Social Survey of 2003 (CGSS)
administered jointly by the Department of Sociology at Renmin University of China and the
Survey Research Center of the Hong Kong University of Science and Technology. The survey
with the Environmental Module (Hong and Xiao 2007) was only conducted in urban regions in
the mainland of China. The final sample has 5073 urban respondents.
The second dataset comes from the urban statistical yearbook in 2003. The data was
published by the National Bureau of Statistics of China. The data were collected from the city
level. Here, I selected socio-economic development variables and industrial pollution indicators.
Our socio-economic development indicators include GDP per capita, city type and percentages
contributions to the total GDP for each selected city of each of three industrial sectors. Our
industrial pollution indicators consist of industrial waste water emissions, industrial dust
emissions, and industrial sulfur dioxide (SO2) emissions, from each city for which I have the
2003 CGSS data.

81

Since the 2003 CGSS data have corresponding city identification codes for each respondent,
I combined the city level data with CGSS data. The combined data set is our final dataset and its
data structure includes two levels. The first level is from respondent‟s social and economic
characteristics and his or her environmental attitudes and behaviors. The second level is the
macro socio-economic and environmental indicators in the respondent‟s city where they reside.
4.3.1 Variables Measurement: Dependent Variables
In this research, environmental attitudes first worked as dependent variables in the multilevel analysis. Later, they were used as a control variable in modeling the effect of individual and
city level variables on public and private pro-environmental behaviors. Our first psychological
variables include the general environmental attitude measured by the NEP. A complete 15 item
2000 version of the NEP scale was used in the 2003 CGSS. There is no consistent standard
about how many items are used to measure the NEP. A confirmatory factor analysis of the 15
items showed that 9 worded items had low factor loadings. Therefore, I kept 6 items out of 15
items (Cronbach‟s Apha is 0.748). Based on these six items, factor score weight output was used
to create the revised NEP scale (see the Table 4-1).
Table 4-1 Six New Environmental Paradigm Scale Items and Percentage Frequencies and
Factor Score Weight in Index

Question Wording
Six NEP items

SD=Strongly
disagree;
D=Mildly
disagree
U=Unsure; MA=Mildly agree; SA=Strongly agree
SD
D
U
A
SA
Factor
Score
(%)
(%)
(%)
(%)
(%) Weight

1.When humans interfere with nature
it often produces disastrous consequences 2.7
2.Humans are severely abusing the 3.6
environment

4.6
8.9

82

11.1
13.2

31.3
33.9

50.4
40.4

0.14
0.13

Table 4-1 (cont’d) Six New Environmental Paradigm Scale Items and Percentage
Frequencies and Factor Score Weight in Index

Six NEP items
(%)
3.Despite our special abilities humans are
still subject to the law of nature
1.1

(%)

(%)

(%)

5.1

19.3

39.5

35

0.10

4.The earth is like a spaceship with very
limited room and resources
1.7

6.5

26.4

33.1

32.3

0.11

5.The balance of nature is very
delicate and easily upset

6.5

25.1

36.9

29.7

0.13

1.7

(%)

Factor
Score
Weight

6.If things continues on their present
course, we will soon experience a major 2.2
7.1
18.7
32.9 39.1
0.15
ecological catastrophe
Note: Items were recoded so that higher scores show more support for the NEP. Missing data
were recorded as “unsure”
As for public and private environmental behaviors, two behavioral indices will be classified
based on ten survey items. Respondents were asked to answer whether in the past year they had
never, sometimes, or often taken ten different pro-environmental friendly behaviors (see Table 41). Through the results of a preliminary principle component analysis, items 1, 2, 3, 4, and 6,
were combined to create a private environmental index (the “private index”) and items 5,7,8,9,
and 10 were combined to form a public environmental behavior index (the “public index”).
Some scholars suggested that using additive indexes might not be the best way to catch the
measurement of these two kinds of behaviors. Despite it not being the best way, there has to be a
way to compare to the findings of the previous literature which use additive indexes to present
these two types of environmental behaviors. For example, Xiao and Hong (2010) classified ten
items and combined each of those ten items to create a public environmental behavior index and
a private environmental behavior index.

83

Table 4-2 Ten Items of Environmental Behaviors and Percentage Frequencies
Never Sometimes Often
(%)
(%)
(%)
Question wording
Private Environmental Behavior
Classifying rubbish
Discuss environmental issues with relatives and friends
Bring your own shopping bags to grocery stores
Recycling plastic shopping bags
Actively pay attention to environmental problems and
information in the media
Public Environmental Behavior
Donating money for environmental protection
Actively participate in environmental campaigns
sponsored by government and workplace
Actively participate in environmental protection activities
held by non-governmental organization
Maintain public woods and grasslands at your own expenses

62.9
31.6
49.6
29

22
56.8
27.8
24.9

15.1
11.6
22.6
46.1

23

45.3

31.7

69.1

27.7

3.2

58.7

30.4

10.9

76.1
82

18.9
12.4

5
5.6

Participate in appeal and express grievances about
2.6
environmental problems.
82.6
14.8
Missing data were minimal from 0.2 % to 0.4%and recoded as the mid-point, “sometimes”
4.3.2 Independent variables
Our socio-demographic variables include age (continuous variable), gender (female=0 vs.
male=1), education (continuous variable), and income (continuous variable
There are two attitudinal variables in this research: environmental attitudes and perceived
local environmental problems. As mentioned above, we will first examine the effect of individual
and contextual level variables on the environmental attitudes measured by the NEP scale with 6
items. Later, the environmental attitudes will be used as one of individual level variables in multilevel analysis of two kinds of pro-environmental behaviors. The second attitudinal variable is
called the perceived local environmental problems. A confirmatory analysis shows that 3 items
out of 10 items have a low factor loading, so I kept 7 out of 10 items (Cronbach‟s Apha is 0.784).
It asks respondents if their living communities have the following environmental problems: air
84

pollution, water pollution, noise pollution, industrial waste pollution, urban garbage pollution,
green space deficiency and destruction of natural vegetation in forests. Based on these seven
items, I used factor score weight output to create the perception of local environmental problems
(For the exact wording, see table 3).
Table 4-3 Seven Items of the Perceived Local Environmental Problems and Percentage
Frequencies
1=Not serious at all; 2=Not very
serious;3=Unsure;4=Somewhat serious,
Question wording
5=Very serious
1(%) 2(%)
3(%)
4(%)
5(%)
1. Air pollution
3.8 16.8
32.3
28.9
18.1
2. Water pollution
4.3 18.6
35.9
26.3
14.8
3. Noise pollution
3.9 16.8
35.5
27.2
16.6
4. Industrial waste pollution
6.2 16.9
46.6
18.6
11.6
5. Urban garbage pollution
4.4 17.1
33.1
28.8
16.7
6. Green space deficiency
4.3 12.5
43.3
24.9
15.0
7. Destruction of natural vegetation in forests
4.1 10.2
63.9
13.9
7.9
Missing data and those who answered “don‟t know” or “don‟t exist” are recoded as “unsure”
Contextual- Level Variables
City economic development level is measured by three variables: GDP per capita, industry
structure (the percentage of the total GDP generated from the primary sector) and city size.
These three indicators were selected to represent city economic development level based on the
following reasons. First, GDP per capita is often considered an indicator of a country‟s standard
of living and the related literature (e.g. Franzen, 2003; Gelissen, 2007) also used the GDP per
capita to test the effect of wealth in certain areas of environmental concern and support for
environmental protection. Second, the industry classification system is used by the government
usually divides industry into three sectors. The primary sector of industry includes agriculture,
mining and raw material extraction. The secondary sector of industry is manufacturing. The third
sector of industry is service production. Some political scientists (e.g. Inglehart, 1990) argued
85

that environmentalism stems from the emergence of post-materialist value among the member of
a new class and most of them holding jobs in the service and information sector instead of the
traditional extraction and industrial sectors. As income per capita rises, agriculture loses its
primacy, giving way first to a rise in the industrial sector, then to rise in the service sector. These
two shifts are called industrialization and post industrialization. China had been an agricultural
country in the past centuries. Therefore, using the percentage of the total GDP generated from
the primary sector (% GDP from the primary sector) catches the essentials of Chinese economic
growth and helps to examine whether decreasing the primary sector contributes to
environmentalism.
City type is an ordinal variable (small and midsize city =1, capital city level=2, large city
=3). In the Chinese society, due to its population size, city type is direct related to all kinds of
resource allocation controlled by the government. Large city here refers to a municipality
directly under the central government and its population size is more than 15 millions. Capital
city is the capital of each province. When the city type moves up a level, its population generally
multiplies by about three. In addition, our previous analysis in chapter 3 showed that the type of
residency is significant to perceived local problems factors and pro-environmental behaviors in
the previous analysis
Industrial pollution is measured by the amount of industrial waste water emission, industrial
sulfur dioxide (SO2), and industrial dust respectively. These three variables are chosen due to
their accessibility in the city statistical yearbook. In China, many pollution indicators are not
publicly released and kept as top secrets by the government. (See Table 4-4 and Table 4-5 for
their correlation matrix and descriptive tables respectively).

86

Table 4-4 Descriptions of City Level Variables
City level variable
City size

City number=72
1=mid size city；
2=capital city;3=large city;
GDP per capita
Continuous ( Yuan)
%GDP from the primary sector Continuous
Industrial wastewater
Continuous (Ten thousand tons)
Industrial Sulfur dioxide
Continuous (Ten thousand tons)
Industrial dust
Continuous (Ten thousand tons)

Mean
1.32

S.D
0.58

18559.57
16.08
12025.05
36283.34
36792.59

11775.44
9.61
15043.86
73701.82
40820.01

From Table 4-4, we can find that economic development is uneven in urban areas in China.
The highest GDP per capita in Chinese cities is 55280 in the coastal city, while the lowest GDP
per capita is 3371 in the interior of China, which also reflects in their percentage of the total
GDP generated from the primary sector. The higher GDP per capita a city is, the lower first
industry a city is.

From the Table 4-5, GDP per capita is negatively correlated with the

percentage of the total GDP generated from the primary sector and positively correlated with city
size. As for three indicators of industrial pollution, industrial wastewater is positively associated
with industrial sulfur dioxide and industrial dust. It should be noted that GDP per capita has no
significant relationship with three indicators of industrial pollution. However, the higher level
city size is, the higher some industrial pollution indicators. The possible reason might be the
effect of pollution size and industrial scale.
Table 4-5 City Level Variables Correlation Matrix
City
size
1

City size
GDP
per capita
0.301*
%GDP
from
the
primary
sector
-0.521**

GDP
per capita

First
Industrial
Industry rate Wastewater

1

-0.654**

1

87

Industrial
Industrial
Sulfur Dioxide Dust

Table 4-5 (cont’d) City Level Variables Correlation Matrix
City
size

GDP
per capita

First
Industrial
Industry rate Wastewater

Industrial
Industrial
Sulfur Dioxide Dust

Industrial
wastewater 0.556**
0.219
-0.289*
1
Industrial
sulfur
dioxide
0.270*
-0.018
-0.159
0.428**
1
Industrial
dust
0.220
0.043
-0.155
0.269*
0.098
1
Note: %GDP from the primary sector is the percentage of the total GDP generated from the
primary sector since China is a developing country and 47% of labors are employed in this
industry. GDP per capita was measured in Chinese yuan. Industrial wastewater was measured in
ten thousand tons, and SO2 and dust was measured in tons. *p<0.1, **p<0.05, ***p<0.01
4.5 Analysis Strategy
A multilevel analysis will be applied to test the hypotheses above. The hierarchical linear
model (HLM) has the capacity to break down the influence of individual-level and municipallevel factors on the outcome variable (Raudenbush & Bryk, 2002). HLM models were run to
predict environmental attitudes and two types of environmental behaviors using HLM software
(Raudenbush, Bryk, & Congdon, 2005). A two-level hierarchical linear model (HLM) was run to
reveal the influences of the individual and contextual factors on environmental attitude and two
types of environmental behaviors. The level 1 model specifies the influence of individual factors
including socio-demographic and psychological variables. The level 2 model tests the effects of
measures of city economic development and pollution level on the individual growth parameters,
when controlling for the effects of social demographic and psychological characteristics. Due to
the different measurement units of dependent variables, standardizing dependent variables was
applied before running the analysis.
This analysis will follow several steps. First, a null model (Model1) was estimated. That is a
model without any explanatory individual and contextual variables.
88

Here, I modeled an

individual‟s environmental attitudes and two types of environmental behaviors (Private vs.
Public) as Yij follows.
Individual level: Yij = β0j+rij
Contextual level: β0j = γ00+u0j
Here, Yij is the dependent variable and represents certain type of environmental attitudes and
behaviors for the responder i in j city. β0j is mean environmental attitudes and behaviors for city
2

j and rij refers to the usual residual error term. The variance of rij is equal to σ , which
represents within- city variability in environmental attitudes and behaviors. γ00 refers to the
grand mean of environmental attitudes and behaviors. The variance of u0j is equal to τ00, which
represents the between city variance in environmental attitudes and behaviors. The aim of this
model is to decompose how much variance there was in the individual level and city level. Intra2

class Correlation Coefficient (ICC) is defined as τ00/ (τ00+ σ ) that gives the proportion of the
total variance in the dependent variable that exists among cities. A low value indicates that there
is little variance among cities.
After this null model, the hypotheses were systematically tested in the following models. To
examine the individual level determinants of environmental attitudes and behaviors, I included
all the individual-level variables with their fixed regression slopes in Model 2; however, I did not
add the city level variables in Model 2. To test the affluence hypothesis, I added city economic
development level variables, such as, GDP per capita, GDP % from the primary sector, and city
size in Model 3.

89

In Model 4, the final model, all contextual level variables are included. There are several
options for model selection. Here, I chose the random intercept model, because the intercept in
this model can be regarded as city level mean of environmental attitudes or behaviors and is
allowed to vary. In addition, individual level coefficients are assumed to be fixed and the random
intercepts is the only random “group effect” (Gelissen, 2007). Therefore, the complete model
with individual- and contextual- level variables can be represented as following:
Individual level:
When Yij represents environmental attitudes for the responder i in j city,
Yij =

β0j

+

β1j

(Gender) +

β2j

(Age) +

β3j

(Income) +

β4j

(Education) +

β5j

(Perceptions of local environmental problems) + rij
When Yij represents two types of environmental behaviors for the responder i in j city,
Yij =

β0j + β1j (Gender) + β2j (Age) + β3j (Income) + β4j (Education) + β5j (NEP) +

β6j (Perceptions of local environmental problems) + rij
Municipal level:

β0j = γ00 + γ01 (GDP per capita) + γ02 (The First Industry Rate) + γ03 (City size) + γ04
(Industrial wastewater) + γ05 (Industrial Sulfur dioxide) + γ06 (Industrial dust) + u0j

β1j = γ10
β2j = γ20
β3j = γ30
β4j = γ40
90

β5j = γ50
β6j = γ60
These equations emphasize that the slope coefficients are fixed. In the full model,

β0j

presents the effects of city level variables on the mean level of environmental attitudes and
behavior on the jth city while the regression coefficients are assumed to have no variation across
units. HLM (Version 6.08) was used here to analyze the data.

Full maximum likelihood was

used here to estimate parameters.
4.6 Results
The results regarding individual level and municipal level effects on environmental attitudes
and private and public environmentally friendly behavior are presented in Table 4- 6, Table 4-7
and Table 4-8, respectively.
Multi-level Analysis of Environmental Attitudes
Model 1 in table 4-6 shows a two level-random intercept null model. I examine how much of
the total variance can be attributed to the individual level and how much to the contextual level.
The points estimate for the grand-mean level of environmental attitude is 23.87. The variance
among cities (0.807) is much smaller than the variance among individuals within cities (14.56).
The chi-square test of the estimate between cities variance component proved to be highly
significant (Chi-square=342.53, DF=71).
For Model 2 in Table 4-6, five individual level variables were added. Some findings are
consistent with previous research. For example, education is the strongest predictor of
environmental attitudes among social demographic variables. Age, income and perceived
environmental problems are positively related to environmental attitudes. Different from some

91

results in the western research, this study found that males have more pro-environmental
attitudes than female.
In Model 3, measures of economic development variables are added in the model. When I
controlled the individual level variables, the percentage of the total GDP generated from the
primary sector is significant negatively related to pro-environmental attitudes while the effects of
the other two variables (GDP per capita and city type) are not significant. An area with a high
percentage of the total GDP generated from the primary sector means the area is underdeveloped. From the perspective of proportion reduction in variance, adding these three macroeconomic variables can help explain 4% of city level variance.
In the final model of table 4-6, three variables used to measure industrial pollution are
added. The percentage of the total GDP generated from the primary sector is still statistically
significant related to environmental attitudes while none of these three industrial pollution
indicators are significant associated to environmental attitudes. Regarding proportion reduction
in variance, city level variance in Model 4 is even higher than in Model 3. This suggests that
these three industrial pollution indictors do not explain any variance in the city level variance.
Table 4-6 Multi-level Analysis of Environmental Attitudes
Model
Fix effect

Model 1
Coefficient
(S.E)

Model 2
Coefficient
(S.E)

Model 3
Coefficient
(S.E)

Model 4
Coefficient
(S.E)

Intercept

23.875***
(0.123)

23.990***
(0.108)

23.977***
(0.105)

23.97***
(0.109)

0.207***
(0.049)
1.087***
(0.083)
0.086**
(0.041)

0.202***
(0.049)
1.082***
(0.084)
0.083**
(0.04)

0.201***
(0.049)
1.082***
(0.084)
0.082**
(0.04)

Individual level Variable
Age
Education
Income
92

Table 4-6(cont’d) Multi-level Analysis of Environmental Attitudes
Model
Model 1
Model 2
0.311***
Male
(0.071)
0.339***
Perceived local Problems
(0.066)
Municipal level variable

GDP per capita
%GDP from the
Primary sector

Model 4
0.313***
(0.072)
0.338**
(0.066)

-0.054
(0.111)
-0.081
(0.153)
-0.302**
(0.120)

City Type

Model 3
0.313***
(0.072)
0.337**
(0.066)

-0.009
(0.114)
-0.094
(0.156)
-0.317**
(0.124)
-0.025
(0.1)
-0.151
(0.084)
-0.02
(0.082)
Variance
Component
2
(χ )
0.572
(249.56)
13.362

Industrial Waste Water
Industrial Sulfur Dioxide
Industrial dust
Variance
Component
Random Effect
2
(χ )
0.807
EFB average level in cities
(342.53)
Level 1 effect
14.656
Note:*p<0.1, **p<0.05, ***p<0.01

Variance
Component
2
(χ )
0.568
(274.192)
13.364

Variance
Component
2
(χ )
0.545
(254)
13.365

Multi-level Analysis of Private Environmental Behaviors
Model 1 in Table 4-7 represents a two-level random intercept null model. I investigated how
much of the total variance can be attributed to the individual level and how much to the
contextual level. The point estimate for the grand-mean level of PEF is 9.033. The variance
among cities (0.289) is much smaller than the variance among individuals within cities (4.373).
This is also reflected in the value of ICC, which is 0.289/ (0.289+4.373) =0.058. This means that
about 6% of the total variance is among cities. I also can find that a chi-square test of the
estimated between cities variance component proved to be highly significant (chi-square=656.67,
DF=71).
93

Model 2 in Table 4-7 examines the extent to which individual-level explanatory variables
are related to private environmental behaviors. Most of the findings fit with our expectations and
all of the findings reported below are statistically significant at p < .001.Older people have more
private environmental behavior than younger ones (b=0.117). NEP has a positive impact on
private environmental behavior. Compared to other determinants, education has the strongest
effect on private environmental behavior. The more educated people are, the more likely they
are to conduct private environmental behaviors. This result also shows that people who perceived
more local environmental problems are more likely to convert their perceptions into private
environmentally friendly behaviors. Although some studies (Aoyogi-Usui et al., 2008) reported
significantly higher participation in environmental behaviors for men, the study here found that
males conducted fewer environmental friendly behaviors in the private sphere (b=-0.156).
Contrary to my expectations, income did not have a significant effect on private environmental
behaviors.
In Model 2 (Table 4-7), these individual-level variables are included, the individual level
variance decreased from 4.373 to 3.893 and the municipal level variance also decreased from
0.289 to 0.178 (compared to Model 1). This amounts to a proportion of explained variances of
10.9% at the individual level and 38.4% at the municipal level. Therefore, there is a noticeable
evidence of a compositional effect. According to views of some scholars (Diez Roux, 2002;
Gelissen, 2007), compositional effect occurs when individual characteristics explain, to some
extent, a person‟s environmental private behaviors. At the same time, if these individual
characteristics are unequally distributed among cities, they also explain some differences in
environmental private behavior among across cities. Since the added individual level variables

94

also explain some part of the municipal level variance, I believe that individual characteristics
may influence the differences of the averages of these cities on environmental private behaviors.
In Models 3 to 4 in Table 4-7, we investigate how macro economic development and
pollution indicators relate to private environmental behaviors. The results show that both
national wealth measured by GDP per capita and percentage of the total GDP generated from the
primary sector have no significant effect on private environmental behaviors. However, city size
has a positive significant association with private environmental behavior. People living in the
big cities are more likely to conduct private environmental behaviors. When macro economic
development indicators were added, the individual level variance kept the same. At the same
time, municipal level variance decreased from 0.178 to 0.146, compared to Model 2. Model 4 in
Table 5 also allows us to examine the relationship between private environmental behaviors and
municipal level pollution indicator. The results show that none of these three pollution
indicators, which include industrial waste water, industrial dust and industrial sulfur dioxide,
have significant relationship with private environmental behaviors.

Table 4-7 Multi-level Analysis of Private Environmental Behavior
Model
Fixed effect

Intercept

Model 1
Coefficient
(S.E)

Model 2
Coefficient
(S.E)

9.033***
(0.072)

Model 3
Coefficient
(S.E)

Model 4
Coefficient
(S.E)

9.11***
(0.059)

4.793***
(0.257)

4.798***
(0.258)

0.117***
(0.032)
0.57***
(0.03)
0.037
(0.028)

0.114***
(0.03)
0.563***
(0.03)
0.033
(0.028)

0.111***
(0.03)
0.563***
(0.03)
0.033
(0.028)

Individual level Variable
Age
Education
Income
95

Table 4-7 (cont’d) Multi-level Analysis of Private Environmental Behavior
Model
Model 1
Model 2
Model 3
-0.156***
-0.153***
Male
(0.03)
(0.03)
0.352***
0.35***
NEP
(0.03)
(0.03)
0.088**
0.082**
Perceived local Problems
(0.034)
(0.034)
Municipal level variable
0.131**
City Type
(0.069)
0.046
GDP per capita
(0.07)
GDP % from the
-0.047
Primary sector
(0.085)
Industrial Waste Water
Industrial Sulfur Dioxide
Industrial dust
Random Effect

EFB average level in cities
Level 1 effect
Note:*p<0.1, **p<0.05, ***p<0.01

Variance
Component
2
(χ )
0.289
(656.67)
4.373

Variance
Component
2
(χ )
0.178
(458.04)
3.893

Model 4
-0.153***
(0.03)
0.35***
(0.03)
0.08**
(0.034)

0.143*
(0.085)
0.062
(0.073)
-0.036
(0.088)
-0.053
(0.052)
0.02
(0.028)
0.042
(0.037)
Variance
Variance
Component Component
2
2
(χ )
(χ )
0.146
0.154
(283.16)
3.894
3.894

Multi-level Analysis of Public Environmental Behaviors
In the analyses of public environmental behaviors, the point estimate for the grand-mean
level of private environmental behaviors is 6.493. It is lower than the point estimate for the
grand-mean level of private environmental behaviors. The variance among cities (0.182) also
turns out to be much smaller than the variance among individuals within cities (3.468). The value
of ICC in public environmental behavior is 0.182/ (0.182+3.468) =0.049, which means that
about 5% of the total variance is among cities. A chi-square test of the estimated between cities

96

variance component is found to be highly significant (chi-square=326.56, DF=71). The
significant chi-square tests of the estimated between-cities variance component in private and
public environmental behaviors indicates that significant variation among cities in these two
kinds of environmental friendly behaviors can be explained by contextual and individual
characteristics.
Model 2 in Table 4-8 examined whether these individual level characteristics such as, age,
income, and gender, which are associated with private environmental behavior have a similar
association with public environmental behavior. The findings are mixed. Income still does not
have any effect on public environmental behavior. Education, NEP and perceived local problems
factors have a positive association with public environmental behaviors. Education is the
strongest predictor in pubic environmental behaviors.
Some results here tell a different story. Different from private environmental behaviors,
younger people play a more active role in the public environmental behaviors (b=-0.006) and
males are more likely to participate in public environmental behaviors (b=0.005). Like Model 2
in Table 4-7, there is also a compositional effect in public environmental behaviors. Compared to
Model 1, with added individual level variables, I found that the variance for individual level
variables decreased from 3.468 to 3.264 and variance for municipal level decreased from 0.182
to 0.165, equal to a proportion of explained variances of 5.8 % at the individual level and 9.3 %
at the municipal level.
In terms of public environmental behaviors, Models 3 and 4 in Table 4-8 tell a different
story. When macro-economic development indicators are added, that municipal variance
decreased from 0.165 to 0.147 while individual level variance is almost same. Different from
what we found in private environmental behavior, Model 3 in Table 4-8 indicates that city type

97

does not have a significant effect on public environmental behaviors. It is worth noting here that
the percentage of the total GDP generated from the primary sector has a significant positive
association with public environmental behaviors. GDP per capita, is found to be not significantly
associated with public environmental behaviors. Regarding the effect of objective environmental
problems, like our previous findings in private environmental behaviors, no evidence is found for
the hypothesis that higher levels of environmental problems are positively related to public
environmental behaviors.
Table 4-8 Multi-level Analysis of Public Environmental Behaviors
Model
Fix effect

Intercept

Model1
Coefficient
(S.E)

Model2
Coefficient
(S.E)

Model4
Coefficient
(S.E)

6.531***
(0.056)

6.535***
(0.054)

6.535***
(0.054)

-0.062*
(0.035)
0.374***
(0.029)
0.004
(0.02)
0.006
(0.034)
0.09***
(0.027)
0.181***
(0.027)

-0.06**
(0.035)
0.375***
(0.03)
0.005
(0.02)
0.005
(0.034)
0.09***
(0.027)
0.18***
(0.027)

-0.06*
(0.035)
0.375***
(0.03)
0.005
(0.02)
0.005
(0.034)
0.09***
(0.027)
0.181***
(0.027)

0.086
(0.056)
0.053
(0.084)
0.211**
(0.098)

6.493***
(0.059)

Model3
Coefficient
(S.E)

0.080
(0.064)
0.048
(0.087)
0.206**
(0.098)
0.022
(0.044)

Individual level Variable
Age
Education
Income
Male
NEP
Perceived local Problems
Municipal level variable
City Type
GDP per capita
GDP % from
Primary sector
Industrial Waste Water
98

Table 4-8 (cont’d) Multi-level Analysis of Public Environmental Behavior
Model
Model1
Model2
Model3
Industrial Sulfur Dioxide
Industrial dust
Random Effect

EFB average level in cities
Level 1 effect
Note:*p<0.1, **p<0.05, ***p<0.01

Variance
Component
2
(χ )
0.182
(326.56)
3.468

Variance
Component
2
(χ )
0.165
(296.39)
3.264

Variance
Component
2
(χ )
0.147
(326.56)
3.265

Model4
-0.015
(0.035)
-0.014
(0.048)
Variance
Component
2
(χ )
0.156
(326.56)
3.265

4.7 Discussions and Conclusion
This study attempts to identify how and to what extent individual level and municipal level
variables are related to environmental attitudes and private vs. public environmental behaviors
through the HLM and finds the existence of compositional effects, which have been examined in
the Western literature (Gelisseon, 2007). As expected, only a small portion of variance (5%) can
be explained by municipal level variables either in public environmental behavior or private
environmental behavior. However, several important conclusions arise from the analysis.
Our findings concerning the individual level variables are largely congruent with the findings
by earlier studies: education, the NEP factor and the perceived local problems factor have a
positive association with private environmental behaviors as well as public environmental
behaviors. Education has the strongest effect on environmental attitude, private environmental
behaviors and public environmental behaviors. It is interesting to find that the NEP factor has the
second strongest effect on private environmental behavior while perceived environmental
problems shows the second strongest effect on public environmental behavior. Does this mean
that NEP, which measures people‟s general environmental attitude, drives individuals to practice
more private environmental behaviors while public environmental behavior is driven more by
99

perceived environmental problems? Since our data is cross-sectional, a longitudinal analysis is
needed to confirm these two driving sources.
Second, our results indicate that women in urban China have higher levels of participation in
private environmental behaviors, while men in urban China have higher levels of environmental
attitudes measured by the NEP scale. As for public environmental behaviors, there are no
significant gender differences. This conclusion is consistent with other findings in both the West
and China (Xiao and Hong, 2010; Tindall et al., 2003). Since most private environmental
behaviors are undertaken in the context of daily routines and since women share greater share of
daily routines, such findings are not surprising. A possible explanations for why men seems to
have higher levels of public environmental behaviors than women is that women spend more
time in their daily routines, and men spend more time outside of housework. This might relate to
traditional gender socialization. Men‟s work is in charge of making money to support family
while female is responsible for doing housework. However, since I do not have additional
information about hours of housework done by urban Chinese respondents, this explanation
cannot be confirmed in this research.
Third, our results suggest that being older in urban China results in a high level of
environmental attitudes and increase participation in private environmental behaviors while
being younger in urban China increase the possibility of participation in public environmental
behaviors. Since the coefficient of age is extremely small, it is a further indication that age has a
weak or no association with environmental behaviors. This finding is also consistent with results
commonly found in the Western settings: age explains a small part of variance of environmental
concern and behaviors.

100

To my surprise, compared to other individual-level socio-economic variables, the influence
of income on environmental behavior is almost negligible. This conclusion is different from the
findings by Guerin et al (2001). In their multilevel analysis of predictors of conservation
behavior in the European Countries, they found that income is a statistically significant
determinant of recycling behavior.
Fourth, regarding the municipal level section of the analysis, our research partially supports
the conclusion that private environmental behavior is greatly influenced by the type of city in
which the individual lives. It is surprising to know that, except for city type, the municipal level
variables, and GDP per capita do not have a significant effect on environmental attitude and two
types of environmental behaviors. However, the percentage of the total GDP generated from the
primary sector does have a significant negatively effect on city‟s mean level of environmental
attitudes. This finding indirectly verified the increasing environmental concern among members
of “a new class” who hold their job in the service and information sector instead of traditional
sector such as agriculture. At the same time, the percentage of the total GDP generated from the
primary sector have a significant positively effect on public environmental participation, it
suggests that Chinese urban residents in high percentage of the total GDP areas are more willing
to participate public environmental behaviors than those in the area with low percentage of the
total GDP generated from the primary sector.
In the Chinese resource allocation system, large cities, such as Beijing and Shanghai, can
acquire more resources from the central government. Since city type in some degree represents
the degree of municipal wealth of each city in urban China. Table 4-7 indicates that city type is
positively significant relate to private environmental behaviors while Table 4-8 shows that city
type has no significant effect on public environmental behaviors. Therefore, the affluence

101

hypothesis is only partially supported by this research. This result is inconsistent with common
findings in the Western setting (Gelissen, 2003; Franzen and Meyer, 2009). For example,
Gelissen (2007) found that wealth measured by GDP per capita is directly related to support for
environmental protection. Franzen and Meyer (2009) found that GDP per capita has a positive
relationship with public concern for the state of the natural environment. However, how much of
this concern or support for environmental protection can be converted into environmental
friendly behaviors is still unknown.
Finally, our results do not support the objective problem hypothesis. Some western scholars
have argued that areas with severe environmental problems have more support for taking action
to solve the problems (Inglehart 1995; Johnson et al. 2005; Marquart-Pyatt 2007). However, in
our data, three indicators of objective environmental problems are not significantly related to
environmental attitude and private or public environmental behaviors.
However, caution is called for when applying measures of objective environmental
problems. Inglehart‟s hypothesis emphasized the importance of local environmental problems.
Here, the objective environmental problems measured mainly use industrial pollution indicators.
These industrial pollution indicators were collected based on each city. The city ranking system
in China is different from in the U.S. In China, the boundaries of cities include counties, which
are comprised of towns and villages. Therefore, cities in China are too large areas to validity
represent the local problems that each respondent is experiencing. Industrial pollution indicators
belong to regional level of environmental problems in the city. Therefore, they cannot adequately
represent the local environmental problems that individuals may confront in their daily life. For
example, urban residents might experience chemical pollutions from nearby factories in their
living community. As Gelissen (2007) pointed out, it is necessary to seek more valid indicators

102

of local environmental problems to provide a stricter test of the objective problems hypothesis in
the future studies.
To my knowledge, this is the first time that a multi-level analysis of environmental attitude
and behaviors in urban China has been conducted. Our findings indicate that the effects of most
individual socio-economic and psychological variables follow a similar pattern as in the West.
As for municipal level variables, our analysis shows that the affluence hypothesis is partially
supported while the objective environmental problesm hypothesis does not get obvious support
here.
These findings have important implications with regard to both theory and policy. It makes
up for the gap in the environmental attitude and behaviors research in the largest developing
countries. In addition, through classifying environmentally friendly behaviors into private vs.
public behaviors, it allows us systematically to explore the determinants of these two kinds of
behaviors. As China becomes more industrialized and urbanized, there is an urgent need for
researchers to understand current Chinese environmental concern and help to promote these
concerns into environmental friendly behaviors and to provide insights for policy makers.
Despite the significance of our findings, several limitations of these data must be considered.
First, in this dataset, private and public environmental behaviors only include five indicators
respectively, which only capture the individual‟s most common environmental behaviors. A
more comprehensive and valid measure of private and public environmental behaviors is needed
in future research. Second, as mentioned before, our measure of objective environmental
problems focuses on regional level environmental problems, instead of local level environmental
problems in the specific community. This may help to provide a more stringent test for the

103

objective problem hypothesis. Also, identifying more comprehensive and valid objective
environmental problems in the specific domain will be necessary in future research.

104

BIBLIOGRAPHY

105

BIBLIOGRAPHY

Bamberg, S. (2006). Is a Residential Relocation a Good opportunity to change Peopl's Travel
Behavior? Results from a theory-driven intervention Study. Environment and Behavior, 38, 820840.
Berenguer, J., Corraliza, J. A., & Martin, R. (2005). Rural-urban differences in environment
concern, attitudes, and actions. European Journal of Psychological Assessment, 21(2), 128-138.
Blankenau, J., Snowden, M., & Langan, M. (2008). Unstanding environmentalism in a red,
agricultural state:The impact of political party identification and place of residence. Sociological
Spectrum, 28(1), 55-80.
Blocker, T. J., & Eckberg, D. L. (1989). Environmental Issues as Women's Issues: General
Concerns and Local Hazards. Social Science Quarterly, 70, 586-593.
Clark, C. F., Kotchen, M. J., & Moore, M. R. (2003). Internal and external influences on proenvironmental behavior: Participation in a green electricity program. Journal of Environmental
Psychology, 23, 237-246.
Cottrell, S. P. (2003). Influence of sociodemographic and environmental attitudes of general
responsible environmental behavior among recreational boaters. Environment and Behavior, 35,
347-375.
Davidson, D. J., & Freudenburg, W. R. (1996). Gender and Environmental risk Concerns.
Environment and Behavior, 28, 302-339.
Diekmann, A., & Franzen., A. (1999). The Wealth of Nations and Environmental Concern.
Environment and Behavior, 31(540-549).
Dietz, T., Kalof, L., & Stern, P. C. (2002). Gender, Values, and Environmentalism. Social
Science Quarterly, 83, 353-364.
Diez Roux, A. V. (2002). A glossary for multilevel analysis. Journal of Epidemiology and
Comunity Health, 56(8), 588-594.
Dolnicar, S., & Grü B. (2009). Environmentally Friendly Behavior: Can Heterogeneity Among
n,
Individuals and Contexts/Environments Be Harvested for Improved Sustainable Management?
Envioronment and Behavior, 41(5), 693-714.
Duan, Y., Luo, Y., & Liu, Y. (1993). Geological Hazards in China. Beijing: Beijing Architecture
and Building Press.
Dunlap, R. E., & Mertig, A. G. (1995). Global Concern for the Environment: Is Affluence a
Prerequisite? Journal of Social Issues, 51, 121-137.
106

Dunlap, R. E., & Van Liere, K. D. (1984). Commitment to the dominant social paradigm and
concern for environmental quality. Social Science Quarterly, 65(1013-1028).
Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). Measuring Endorsement
of the New Ecological Paradigm: A Revised NEP Scale. Journal of Social Issues, 56(3), 425442.
Dunlap, R. E. & York, R. (2008). The globalization of environmental concern and the limits of
the post-materialist explanation: Evidence from four cross-national surveys. Sociological
Quarterly, 49 529-563.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to
theory and research. MA: Addison-Welsey.
Frantz, A., & Meyer, R. (2004). Environmental Attitudes in Cross-national perspective: A
multilevel analysis of the ISSP 1993 and 2000. European Sociological Review, 26(2), 219-234.
Freymeyer, H. Robert, & Johnson, E. Barbara. (2010). A Cross-cultural Investigation of Factors
Influencing Environmental Actions. Sociological Spectrum 30(2), 184-195.
Guerin, D., Crete, J., & Mercier, J. (2001). A Multilevel Analysis of the terminants of Recycling
Behavior in the European Countries. Social Science Research, 30, 195-218.
Haller, M. & Hadler,, M. (2008). Dispositions to Act in Favor of the Environment: Fatalism and
Readiness to Make Sacrifices in a Crosss-National Perspective .Sociological Forum, 23 (2), 281311.
Hayes, B. C. (2001). Gender, Scientific Knowledge, and Attitudes toward the Environment: A
Cross-National Analysis. Political Research Quarterly, 54(3), 657-671.
Hines, J. M., Hungerford, H. R., & Tomera, A. N. (1987). Analysis and synthesis of research on
responsible environmental behavior: A meta-analysis. Journal of Environmental Education,
18(2), 1-8.
Hunter, L. M., Hatch, A., & Johnson, A. (2004). Cross-National Gender Variation in
Environmental Behaviors. Social Science Quarterly, 85(3), 677-694.
Hunter, L. M., Hatch, A., & Johnson, A. (2004). Cross-national gender variation in
environmental behaviors." 85: 677-694. Social Science Quarterly, 85, 677-694.
Inglehart Ronald.(1990). Culture Shift in Advanced Industrial Society. Princeton University
Press.
Inglehart, R. (1995). Public Support for Environmental Protection: Objevtive and Subjective
Values in 43 Societies. Political Science and Politics, 28, 57-72.
107

Johnson, Martin, Paul Brace, and Kevin Arceneaux.(2005). Public Opinion and Dynamic
Representation in the American States: The Case of Environmental Attitudes. Social Science
Quarterly 86:187–108.
Kaiser, F. G., & Gutscher, H. (2003). The Proposition of a General Version of the Theory of
Planned Behavior: Predicting Ecological Behavior. Applied Social Psychology, 33(3), 586-603.
Kaiser, F. G., Hubner, G., & Bogner, F. X. (2005). Contrasting the Theory of Planned Behavior
with the Vaule-Belief-Norm Model in Explaining Conservation Behavior. Journal of Applied
Social Psychology, 35(10), 2150-2170.
Kasapoğlu, M. A., & Ecevit, M. C. (2002). Attitudes and Behavior toward the Environment: The
case of Lake Burdur in Turkey. Environment and Behavior, 34(3), 363-377.
Mobey, C., Vagias, W. M., & Deward, S. L. (2010). Exploring Additional Determinants of
Environmentally Responsible Behavior: The Influence of Environmental Literature and
Environmental Attitudes. Environment and Behavior, 42(4), 420-447.
Oreg, S., & Katz-Gerro, T. (2006). Predicting proenvironmental Behavior cross-nationally:
Value, the Theory of Planned Behavior, and value-belief-norm theory. Environment and
Behavior, 38(4), 462-483.
Ostman, R. E., & Parker, J. L. (1987). Impact of education, age, newspapers and television on
environmental knowledge, concerns and behaviors. Journal of Environmental Education, 19(1),
3-9.
Raudenbush, S. W., & Bryk, A., S. (2002). Hierarchical linear models: Applications and data
analysis methods. 2nd edition. Newbury Park: CA: Sage.
Raudenbush, S. W., Bryk, A., S, & Congdon, R. T. (2005). HLM-6. Lincolnwood, IL: Scientific
Software International.
Marguart-Pyatt, S. T. (2007). Concern for the Environment among General Publics: A Crossnational Study. Society and Natural Resources, 20, 883-898.

Marguart-Pyatt, S. T. (2008). Are There Similar sources of Environmental Concern? Comparing
Industrialized Countries. Social Science Quarterly, 89(5), 1312-1335.
Schultz, P. W., & Oskamp, S. (1996). Effort as a moderator of the attitude-behavior relationship:
General environmental concern and recycling. Social Science Quarterly, 59, 375-383.
Scott, D., & Willits, F. K. (1994). Environmental Attitudes and Behaviors: A Pennsylvania
Survey. Environment and Behavior, 26(2), 239-260.

108

Sherkat, D. E., & Ellison, C. G. (2007). Structuring the religion-environment connection:
Identifying religious influences on environmental concern and activism. Journal of the Scientific
Study of Religion, 46(1), 71-85.
Stern, P. C. (2000). Toward a coherent theory of environmentally significant behavior. Journal
of Social Issues, 56, 407-424.
Stern, P. C., Dietz, T., & Guagnano, G. A. (1995). The New Ecological Paradigm in Socialpsychological Context. Environment and Behavior, 27(6), 723-743.
Stern, P. C., Dietz, T., Kalof, L., & Guagnano, G. A. (1995). Values, beliefs, and
Proenvironmental action:Attitude formation toward emergent attitude objects. Journal of
Applied Social Psychology, 25(18), 1611-1634.
Theodori, G. L., & Luloff, A. E. (2002). Position on Environmental Issues and Engagement in
Proenvironmental Behaviors. Society and Natural Resources, 15, 471-482.
Tindall, D. B., Davies, S., & Mauboules, C. (2003). Activism and concervationbehavior in an
environmental movement: The contradictory effects of gender. Society and Natural Resources,
16, 909-932.
Wang, Y. (2004). Environmental Degradation and Environmental Threats in China.
Environmental Monitoring and Assessment, 70, 161-169.
Xiao, C., & Hong, D. (2010). Gender Differences in Environmental Behaviors in China.
Population and Environment, 32(1), 88-104.
Xiao, C., & McCright, A. M. (2007). Environmental Concern and Socio-demographic Variables:
A Study of Statistical Models. The Journal of Environmental Education, 38(2), 3-12.
Zelezny, L. C., Chua, P.-P., & Aldrich, C. (2000). Elaborating on Gender Differences in
Environmentalism. Journal of Social Issues, 56(3), 443-457.
Zeng, Y., Gu, D., Purser, J., Hoenig, H., & Christakis, N. (2010). Associations of Environmental
Factors with Elderly Health and Mortality in China. American Journal of Public Health, 100(2),
298-304.

109

Chapter 5 Conclusion and Discussion
5.1 Overview
This dissertation research, comprised of three papers, sets out answer three research
questions: 1) Is the measurement of environmental concern in China different from in western
societies? 2) How will the selection of particular measures of environmental concern affect their
relation to social-demographic variables? 3) What are the effects of individual and community
level variables on two types of environmental behaviors?
Using the 2003 General Social Survey (GSS) data and some variables derived from 2003
Urban Statistical Book, I not only try to build a measurement of environmental concern and its
links to the respondent‟s individual social-demographic variables. I also estimate the effects of
individual and community level variables, such as, GDP per capita and regional environmental
pollution index on environmental attitudes and two kinds of environmental behaviors: private
pro-environmental behaviors and public pro-environmental behaviors. This chapter summarizes
the findings from these three papers, outlines policy implications and discusses the future
research directions.
5.2 Summary of Research Findings
In the first paper of chapter 2, through examining the 2003 GSS data, I proposed a
measurement model of environmental concern in China, which includes four dimensions:
Economic Environmental Trade off, the NEP, Pro-environmental behaviors, and Perceived
Locally Environmental Problems. By exploring the correlation pattern among the constructed
four latent factors, I found that Pro-environmental behavior (PEB) factor has a high association
with NEP and Economical Trade-off factor, which verifies the previous findings that the
measurement of environmental concern put more weights on the NEP scale. From the
110

perspective of environmental constructionist, traditional and new rising mass media plays an
important role in contributing new environmental paradigm worldview for the public.
As I mentioned before, previous research did not take perceived local environmental
problem into account in their constructed model. In my measurement model, perceived locally
environmental problems factor‟s relationship with the other factors is contrary to original
expectation: it has no significant association with Economic Environmental Trade Off Factor.
Why do people not to take a more active action on protecting their local environment where
people perceive more locally environmental problems? The answer to this question might require
some insights from social psychological perspectives and political systems existing in the
Chinese society.
In the second paper, I applied the structural equation model to examine how the selection of
particular measure of environmental concern influences social demographic characteristics. The
results suggest that more attention should be paid to the selection of particular measure of
environmental concern. Compared to the similar studies in developed societies, the findings here
has some commonalities as well as differences. For example, regarding the effect of education,
the result exhibits that the more educated person generally has more concern for the
environment, which is parallel with most previous research. However, in terms of gender, the
result here suggests that males have more environmental concern than females in urban China,
which is different from some previous research results.
The significant associations between occupation, education and the pro-environmental
behaviors factor in U.S and China may reflect the fact that environmental quality is usually
treated as a luxury good, and the privilege of upper and middle classes in the Western societies
as well as in China. However, does this conclusion suggest that rich and upper class people in

111

China are more likely or the same as their Western counterparties to conduct proenvironmentally friendly behavior in private field? Since the items used to measure the proenvironmental behavior factor are classified as public environmental behaviors, this question
warrants further research. The third paper of chapter 4 provides some insights to this question.
In the third paper, I used HLM to identify how and to what extent individual level and
municipal level variables are related to private vs. public environmental behaviors. I find the
existence of compositional effects, and it means that both individual level variables and
contextual level variables contribute to explain the total variance. Such effects have been
examined in the Western literature. As expected, only small portion of variance (5%) can be
explained by municipal level variables either in environmental attitudes and public
environmental behavior or private environmental behavior. However, several important
conclusions arise from the analysis. First, in the individual level variables, most of findings are
congruent with previous research in the U.S. and European societies. However, I found that in
China, income has no significant association with these two kinds of pro-environmental friendly
behaviors. In contrast, several scholars (Clark et al. 2003; Gatersleben et al. 2002) found that
people with a high income in western societies are more likely to take part in different proenvironmental friendly activities than people with a low income. The data here demonstrate that
rich people in urban China do not show a significantly high level participation than poorer
people in different pro-environmental behaviors. Does that indicate that rich people in China feel
less social responsibility than their counterparts in the Western society?
In the municipal level, research partially supports the conclusion that although the type of
city has no effect on environmental attitudes, private environmental behavior is greatly
influenced by the type of city in which the individual lives. It is surprising that GDP per capita

112

has no effect on environmental attitude and behaviors while % GDP generated from the primary
sector has a significant effect two kinds of environmental attitude variables: 1) on environmental
attitudes measured by the six item NEP scale and 2) city‟s mean level of participation in public
environmental behaviors. In addition, the results here do not support the objective problem
hypothesis. Three indicators of objective environmental problems are not significantly related to
either environmental attitudes and private or public environmental behaviors.
5.3 Significance and Limitations of the Study
To my knowledge, this is the first comprehensive study to use a national sample to examine
environmental attitudes and pro-environmental behaviors in China. Answers to the raised
research questions in this study meet the urgent need for more evidence of knowledge about
urban residents‟ environmental attitudes and behaviors. As mentioned previously, most extant
studies in this area have focused on developed countries or areas. There is little research
exploring the status of environmental behaviors and attitudes in urban China. As the Chinese
society becomes more industrialized and urbanized, there is a need to increase understanding of
the construction of environmental concern based on Chinese social background and its social and
economic characteristics and further promote these concerns into pro-environmental behaviors.
This research is also designed to advance the existing knowledge about environmental
attitudes and behaviors among common Chinese people that has so far received limited attention.
The findings of this study may be used as the base for future research on how to improve
sustainable life styles among the Chinese people or serve as a reference for studying social and
economical factors influencing environmental attitudes and behaviors among urban Chinese
residents.

113

This study has great policy implications for government policy making in the national level
and global level. Our results demonstrated that environmental concern is a process of multidimension construction. Focusing on one dimension and ignoring the other cannot fully catch the
essentials of environmental concern. To increase environmental concern of the public in China,
government policy makers need to seek the balance between the economic growths and
environmental protection and switch to a more ecological development road. Past experiences
show that it is not an easy task to develop programs that promote ecologically and economically
sound policy.
Our analysis revealed that education plays an important role in contributing to the rising of
environmental concern. Therefore, government and schools should spend more resource in
teaching children about the diversity of natures and cultures. A greater appreciation of the
diversity of environment should help us realize the consequence of environmental degradation.
In addition, since there is gender difference in willingness to engage in public and private
environmental behaviors, it suggests that our environmental education programs should have
specific purposes. For example, these programs should encourage male groups to participate
more in private environmental behaviors and female groups to take part in more public private
environmental behaviors. Finally, due to the different effects of public and private environmental
behaviors, different education or campaigns should be design to encourage these different kinds
of behaviors.
In the global level, concern for environment is not restricted to wealthy nations. It also
spreads well in the developing countries such as, China. However, recent global environmental
protection effects have been stagnated in part due to the actions of a handful of rich nations. For
example, Bush administration refused to sign the Kyoto Protocol. Therefore, it is unfair to blame

114

developing nations for the lack of progress in protecting the global environment. Only by
acknowledging the fact that the citizen‟s livelihoods in developing countries mainly depend on
available natural resources, can effective polices be designed and implemented.
Our analysis also show that there are no huge difference between China and the developed
countries such as the U.S. in terms of measure of environmental concern measured by the six
NEP items. The results showed that the decreasing percentage of traditional sector can contribute
to the increasing environmental concern and public environmental behaviors due to the
formation of “new class”. Since the members of new class are mostly well educated and mainly
work on service sectors, therefore, whether governments in developed countries or developing
countries have obligation to upgrade their industrial sectors and make service sector have more
members working in it.
Despite the significance of our findings, there are limitations in this research due to the data
availability and data quality .First, with the increasing global environmental issues, such as
global warming and global ozone depletion, future research in measure of environmental concern
in developing countries like China, should consider how the public perceives these global
environmental issues. Second, the classification of private and public environmental behaviors
only include five indicators respectively, which only capture the individual‟s most common
environmental behaviors.
Future research might need to identify additional valid items to capture those two kinds of
behaviors. For example, individuals purchasing energy efficient products with more cost should
be added in private environmental behaviors. Third, it should be noted that our data are limited
to urban residents and exclude the residents in rural areas. Therefore, applying the generalized
results to the whole China should be careful.

115

BIBLIOGRAPHY

116

BIBLIOGRAPHY

Clark, C. F., J.Kotchen, M., & Moore, M. R. (2003). Internal and external influences on proenvironmental behaviors: Participation in a green electricity program. Journal of Environmental
Psychology, 23, 237-246.
Gatersleben, B., Steg, L., & Vlek, C. (2002). Measurement and Determinants of Environmentally
Significant Consumer Behavior. Environment and Behavior, 34(3), 335-362.

117