1 An Expanded Perspective on the Policy Implications of Environmentally Related Behaviors 2 William Kilbourne Clemson University Pierre McDonagh Dublin City University Andrea Prothero University College Dublin David Marshall University of Edinburgh Contact Author: William Kilbourne Department of Marketing Clemson University 245 Sirrine Hall Clemson, SC 29634-1325 USA 3 Abstract Environmentally related behaviors have been a topic of study for marketing and public policy researchers for several decades now (for examples see Allen, Calantone, and Schewe 1982; Schwepker and Cornwell 1991; Taschian, Slama, and Taschian 1984). The purpose of this paper was to expand the domain of inquiry by testing a more comprehensive model of such behavior. Drawing from past research on values, the new environmental paradigm, beliefs, concern, and environmental behaviors, a causal model was proposed and tested with a random sample of US adults. The results of the study validate the model indicating that there are policy implications at each stage of the causal sequence leading to environmentally responsible behaviors. Prior work has highlighted the possible positive and negative impact of public policy on marketplace functions (see Brockett and Golden 1990) Key words: Environment, values, environmental concern, NEP, environmentally responsible behavior 4 Introduction For the last 35 years, the study of the environment and people’s effect on it has proceeded at a steady pace. Within that time, researchers have examined the problem from a number of perspectives including, but not limited to, measures of concern (Granzin and Olsen 1991; Schwepker and Cornwell 1991), attitudes (Berger and Corbin 1992; Milfont and Duckitt 2004), environmentally responsible behaviors (ERB) (Berger and Corbin 1992), folk ecological wisdom (Stern 2000), and value orientations (McCarty and Shrum 2001; Stern and Dietz 1994). Among the consistent conclusions drawn is the belief that the processes through which individuals perceive and relate to the natural environment are quite complex (see for example Granzin and Olsen 1991). Among other things, psychological, sociological, economic, technological, religious, and historical components affect the relationship. This suggests that an understanding of the essential processes will require complex models incorporating many of these dimensions. Allen, Calantone, and Schewe (1982), for example, conclude that certain constituencies (say Canadians) may have better understanding of one issue, like energy saving due to government policy, than other groups. Likewise, people of different political orientations have different views on both the existence of environmental problems and their policy implications (Dryzek 1987). Public policy research, therefore, has an integrative role to play in any proposed environmental solutions. As evidence of the importance of environmental policy issues, the Journal of Consumer Research, the Journal of Advertising, and the Journal of Public Policy & Marketing have had special issues relating to the topic. 5 For effective and enduring change toward environmentally responsible behavior (ERB), policy instruments and procedures should take into consideration the multiple levels through which behaviors become manifest. Examining the level of environmental concern and relating it to behaviors may reveal short term effects, but if the antecedents to environmental concern remain unchanged, then the gains may be ephemeral if the antecedents have negative effects on concern itself. Granzin and Olsen (1991) argue that for policy to be effective, a clear understanding of individuals is necessary. Such an understanding implies a more comprehensive examination entailing multiple layers of factors leading to behavior. This is because the antecedent factors left unattended can militate against policy intentions. Alwitt and Pitts (1996) argue that policy initiatives aimed too low in the model of environmental behaviors may be responsible for the apparent attitude-behavior gap found in the US. Berger and Corbin (1992) concur regarding the ability of attitudes to predict ERBs. McCarty and Shrum (2001) argue that attitudes alone are insufficient in predicting ERBs because they have a different character than normal consumption behavior in that immediate self-interest is not the motivating factor. They argue that a more complete spectrum of antecedent variables is necessary to better understand ERB. In expanding the domain of inquiry, Dobscha and Ozanne (2001) characterize the ecological self as a nondominating path to change in ERBs. Matthews (1991) describes the ecological self and indicates that it is much more complex than current research characterizes it. This suggests that treating ERBs as the outcome in a hierarchical model of constructs that lead to behaviors could serve to reduce the attitude-behavior gap by providing the clearer understanding called for by Granzin and Olsen (1991) and Berger 6 and Corbin (1992). Stern (2000) argues that environmental behaviors themselves are more complex than has been considered in the literature, and the domain of inquiry should be expanded there as well. Stern (2000) argues, for example, that there are direct and indirect ERBs, and that these need to be considered as different types in models that attempt to determine the factors that influence such behaviors. Much of the research in marketing has focused on direct behaviors such as recycling, energy consumption, and purchasing green products (Kilbourne and Beckmann 1998). The indirect behaviors include such things as joining environmental organizations and donating money to environmental organizations. Different types of policy initiatives are likely to stimulate each of these types of behavior in different ways and result in different consequences. Responding to the call for more complex models of ERB, Stern, Dietz, and Guagnano (1995) proposed a hierarchical model that incorporates multiple levels of analysis beginning with social structural variables and proceeding through behavior. Ajzen and Fishbein (1977) also suggest that measurement models conjoining different constructs having different levels of abstraction should not use measures that are of different levels of specificity. Thus, relating values measures directly to behaviors is likely to result in poorly specified relationships. The hierarchical model offered by Stern, Dietz, and Guagnano (1995) has this suggestion built into it, and it argues that the hierarchical order proceeds from the more general to the more specific constructs. We take this approach in the present study and develop and test a hierarchical model similar to that proposed by Stern, Dietz, and Guagnano (1995), but with additional levels. 7 The proposed model, shown in Figure 1, begins with the measure of individual values shown to be related to the environment (Stern and Dietz 1994) and then proceeds successively through general ecological beliefs as measured by the new environmental paradigm (NEP) (Cordano, Welcomer, and Scherer 2003; Dunlap and Van Liere 1978), specific environmental beliefs, environmental concern, and then behavior. The hierarchical model has also been supported by Fransson and Garling (1999) who argue that it is necessary because as the specificity of the measures between variables increases, their predictive ability decreases. Arranging constructs from the more general to the more specific might increase predictive validity. Environmental values, for example, are more likely to be predictive of the NEP than they are of environmental concern or behaviors. Likewise, the NEP would be more predictive of specific environmental beliefs than would environmental values. This implies that a causal modeling approach to such hierarchically arranged constructs is important as well (Fransson and Garling 1999). The model hypothesized in this study is such a model following the approach described by Stern, Dietz, and Guagnano (1995). It begins with the more abstract constructs and moves to successively more tangible constructs. The model proposed answers the consistent call for a more thorough examination of the variables antecedent to ERBs. Such a model will provide a better starting point for the development of environmental policies. _______________________________________________________________________ Place Figure 1 about here _______________________________________________________________________ 8 Literature Review Value Orientations Dietz and Stern (1995) argue that values guide choices in deliberative tasks, and that among these choices are those that relate to ERBs. They also argue that values and worldviews are antecedent to specific beliefs that precede behaviors. In their proposed causal model, cultural institutions are at the highest level followed successively by values, worldviews and general beliefs (the NEP), specific beliefs and attitudes, behavioral intentions, and behaviors. Among the possible approaches to the role of values in behavior, and one that has been used frequently in environmental studies, is that of Schwartz (1992) whose model of values is circumplex, indicating that values exist in a circular framework with adjacent values being similar to each other and values across from each other being different. In this system, presented in Figure 2, Schwartz (1992) also identifies four higher order values he describes as self-transcendence, self-enhancement, openness, and conservation (tradition) that are constituted by subsets of the ten individual values making up the circumplex model. _______________________________________________________________________ Place Figure 2 about here _______________________________________________________________________ Stern, Dietz, and Guagnano (1998), Schultz and Zelezny (1998) and Schultz and Zelezny (1999) conclude that both self-enhancement and self-transcendence are 9 predictive of environmental concern. It has also been demonstrated that individuals with transcendent values are more likely to engage in pro-environmental behaviors (Dietz, Stern, and Guagnano 1998), and pro-social values are more evident than are competitive values among those with higher environmental concern (Stern 2000). Stern, Dietz, and Guagnano (1998) confirmed the effects of self-transcendence and self-enhancement, but they also determined that openness to change was marginally related to the NEP. Schultz and Zelezny (1998) also found limited support for a positive environmental effect for openness. Thus, the limited evidence available suggests that transcendence and enhancement values are consistently related to ERBs, albeit in opposite directions, and openness to change is weakly related to environmental concern. Dietz and Stern (1995) argue that this is because the egoistic orientation leads actors to focus on factors that affect themselves, while altruistic actors focus on collective welfare. Because the environment affects both the individual and the collective, one’s value orientation should influence ERBs. Those who are altruistically oriented (self-transcendent) are likely to behave more environmentally responsibly, while those who are egoistic (selfenhancement) are more likely to behave in less environmentally responsible ways as it is in their immediate interest to do so. To change one’s behavior, it is reasonable to argue that they must be open to change. If so, then switching to more ERB is more likely if one is open to change, though, because of the inconsistent empirical results, it is not certain what the antecedents for change have to be. Environmental Beliefs 10 General beliefs and specific beliefs are two dimensions of environmental beliefs. One’s environmental worldview contains these general beliefs about the functioning of the environment. Beliefs about limits to growth, the role of humans in the environment, and the susceptibility of nature to disruption inform the worldview, and the NEP (Cordano, Welcomer, and Scherer 2003; Dunlap and Van Liere 1978) addresses these three constructs. Thus, this type of belief relates to how the individual sees the environment as a whole. These beliefs may be accurate assessments based on environmental knowledge, or they may be the product of what Stern, Dietz and Guagnano (1995) refer to as ecological folk wisdom. The important aspect is that such wisdom represents the background for specific environmental beliefs. General beliefs provide the context for the development of specific beliefs, but they do not relate to any specific problem. Specific environmental beliefs refers to belief in the existence of specific environmental problems such as ozone depletion and global warming. In the proposed causal model, general beliefs about the environment are antecedent to specific beliefs. Stern (2000) refers to these focused beliefs as the belief in adverse consequences to valued objects. It is important to note that the individual need not understand the relationship to draw conclusions about it that direct behavior. For example, Kempton, Boster, and Hartley (1995) demonstrated that US citizens generally believe that global warming and ozone depletion are directly related even though they are not. This suggests that beliefs should be measured on two levels, the general and the specific. The NEP is the measure most often associated with general environmental beliefs, but relevant specific beliefs might change over time as knowledge increases or 11 the attribute becomes more exposed through popular culture (Bechtel, Verdugo, and Pinheiro 1999). We posit here that general environmental beliefs precede specific beliefs, and both are a necessary condition for the development of concern. This is reasonable because, if the individual does not believe that there are environmental problems or believes that nature is sufficiently resilient to remediate itself, then increased concern for the environment is unjustified. Steg and Sievers (2000), for example, demonstrated that people who think nature exists in a precarious balance have higher problem awareness and policy concerns than those who think nature is resilient and stable. This suggests that one’s general view of the environment as reflected in myths of nature (ecological folk wisdom) affects both beliefs about problems and concern for remedial policies. Environmental Concern Fransson and Garling (1999) argue that environmental concern has been described in various ways ranging from attitudinal focus to value orientations. Concern is often synonymous with environmental attitudes and used interchangeably. Stern (1992) relates concern to specific value orientations including altruistic, self-interest, anthropocentric altruism, and egoistic values. Anthropocentric altruism has also been referred to as human welfare ecology (Eckersley 1992; Kilbourne 1995) characterizing it as concern for the environment to the extent that it affects human well-being, but not concern for the environment per se. Concern has also been related to other deep rooted beliefs such as religious beliefs (White 1967) and post-materialist (Inglehart 1981) values. Milfont and Duckitt (2004) also argue that environmental concern has been used to refer to environmental attitudes. They define environmental attitudes according to the 12 tripartite model popular in psychology consisting of affective, cognitive, and conative components as they relate to the environment and argue for a conceptual differentiation between attitudes and concern. This is because researchers use concern as a general attitude, as opposed to specific attitudes, about specific environmental problems. In this paper, the two terms are differentiated, albeit for a different reason. The rationale is that one can hold very clear attitudes and beliefs about environmental problems such as global warming and still be unconcerned about them. This is because attitudes, beliefs, and knowledge may be necessary conditions for concern, but they are not sufficient conditions. Schultz (2001), for example, argues that Schwartz’s (1968) norm activation model indicates that concerns will become activated when there is a threat to a personal value or a valued object. If one does not value the environment or some aspect of it, then knowledge about its condition would have no effect. Thus, in the absence of beliefs or knowledge about environmental problems, there is no need for concern. Before the norm activation process can engage, some cognitive aspects of the problem must be in evidence. Beliefs are not only different from concerns, but they are also prior to concern in the causal order. Environmental Behavior Stern (2000) categorizes environmentally significant behaviors into two basic types. He refers to the two types as direct and indirect actions. Direct actions are those elicited by the individual that have an immediate effect, however small, on the environment. Recycling, reusing, and green purchasing fall in this category, and they are only effective in minimizing environmental problems if large numbers of people use 13 them. Indirect behaviors are those that may have environmental consequences in the future and on a more general scale. These include, inter alia, donations to environmental organizations, reading environmental magazines, and contacting policy makers. Dietz, Stern, and Guagnano (1998) argue that these types of behaviors ought to be distinguished from each other. Stern (2000) also refers to these types of behavior as public and private sphere behavior and concurs with the notion that the former has indirect effects and the latter has direct effects that materialize in the aggregate when many individuals elicit the same behaviors. The proposed model incorporates direct and indirect behaviors, and measures of each are used. Structural Model The structural model examined in this paper consists of the four areas described above; values, beliefs, concern, and behavior. The causal sequence is a modification of the model proposed by Stern, Dietz, and Guagnano (1995). The original model proposed that the causal sequence was, from beginning to end, social structural factors, values, general beliefs (NEP), specific beliefs, behavioral intentions, and behavior. Johnson, Bowker, and Cordell (2004) proposed an abbreviated model that consisted of the social structural, general beliefs, and behavior. The model proposed here expands on their abbreviated model but does not include the social structural or behavioral intention components of the Stern model. The hypotheses generated by the proposed model are as follows: H1. The second order values construct is positively related to general environmental beliefs as measured by the second order NEP construct. 14 H2. The second order NEP construct is positively related to specific beliefs. H3. Specific environmental beliefs are positively related to environmental concern. H4. Environmental concern is positively related to both direct and indirect ERBs. Methodology Sample The data collection procedure for the study was a telephone survey using a sampling frame provided by Scientific Telephone Samples, a research service that provides random samples of telephone numbers throughout the US. The callers eliminated refusals and replaced no-answers into the database where they were available for a successive selection. A no-answer was replaced 9 times before removed. Interviewers repeated the process until they finished the required number of interviews. The sample in this study consisted of 337 respondents selected using a random sample of US residents. The sample was restricted to those who were 18 or more years old. Data were collected using a telephone survey and the average interview time was 25 minutes per completed questionnaire. Thirty-four questionnaires were eliminated because of large amounts of incomplete data that could not be imputed. The final sample consisted of 303 respondents of which 55% were female. Median family income was approximately $45,000, median and average age were both 48 years, the median education level fell in the category “some college,” and 57% of the sample were married. These sample characteristics parallel the US population closely when considering only those who are 18 or older. 15 Measurement Instrument The constructs used were in this study were an abbreviated Schwartz’s (1994) value inventory (SVI), an eight item environmental concern scale adapted from Kilbourne, Beckman, and Thelen (2002), an eight item measure of specific environmental beliefs (ENB) followed by a reduced new environmental paradigm (NEP) scale taken from Dunlap and Van Liere’s (1978) original scale. The last scale was a 10 item proenvironmental behavior scale consisting of yes or no responses to behavioral questions. The final seven questions were demographic. Because of the large number of items in the original scales and the restrictions the telephone survey imposes on the length of the questionnaire administered, these construct measures were reduced by eliminating items from the original scales. Richins (2004), Cordano, Welcomer, and Scherer (2003), and Stern, Dietz, and Guagnano (1998) all support the use of such abbreviated scales in research including multiple constructs. The short scales are necessary to keep the overall length of questionnaires to a manageable level. Because the measurement constructs used in the present study were new or shortened versions of the originals that have not been adequately tested, all the constructs used in the model were subjected to confirmatory factor analysis (CFA) before the structural model was tested. The fit criteria used were those suggested by Hair et al. (1998), and include the comparative fit index (CFI) (>.90), the incremental fit index (IFI) (>.90), and the root mean square error of approximation (RMSEA) (< .08). Because some of the constructs have a restricted number of indicator items, Byrne’s (2001) procedures were followed. She suggests that, in cases with too few degrees of freedom, parameters 16 shown to be statistically equivalent can be constrained to be equal, thus achieving the required positive degrees of freedom. For the SVI, a three-factor construct was used with three items per factor. The three factors used were those shown most frequently to be related to environmental attitudes. These were self-transcendence (ST), self-enhancement (SE), and openness to change (OP). Tradition has been only sporadically related to environmental attitudes and was not used. SE has been shown to be negatively related to environmental attitudes, so its items were reversed to be consistent with the other two factors. The indicator items for the scale consisted of seven of the nine items proposed by Stern, Dietz, and Guagnano (1998) and two additional items, and the three factor model was shown to fit the data very well. The fit statistics were all above .97 and the RMSEA was .045. All factor loadings were above .5 and the average loading was .61. The reliabilities of the individual scales were assessed using Coefficient Alpha and were .71 for ST, .63 for SE, and .56 for OP. These were within the limits suggested by Peter (1979) for constructs in the early stages of development. In addition, the correlations between factors were less than unity indicating discriminant validity, but they were also relatively high suggesting a second order factor structure. The Appendix provides all the indicator items and the statistics for the scales used. The second order structure was imposed on the model and it was retested. The results of this analysis indicated that the second order factor structure was well supported. To achieve positive degrees of freedom in the second order, the error terms for ST and OP were constrained to be equal. This is justified because the parameter comparisons calculated in AMOS (Arbuckle 2005) indicated that they were statistically equivalent 17 (Byrne 2001). Thus, the values construct for the structural model was a second order factor model with three first order factors, ST, SE, and OP. For the NEP construct, a reduced model was used that contained three factors with three indicator items each. The items chosen represented three dimensions of the new NEP scale (Dunlap et al. 2000) and the reduced form of the new NEP (Cordano, Welcomer, and Scherer 2003). The dimensions used were limits to growth, human domination (reverse scored), and balance of nature with three items representing each dimension. This reduced NEP construct was subjected to CFA, and the results indicated a very good fit with all fit statistics above .96 and an RMSEA of .06. All factor loadings were above .56 with an average loading of .69. Coefficient alpha for the dimensions were .77 for balance of nature, .61 for limits to growth, and .80 for human domination. The correlations between factors were also significant indicating a second order factor. The second order factor was imposed on the data and the model retested. The second order model fit the data well with all fit statistics well within the required limits. The next level of the model was environmental beliefs, and the scale consisted of six items derived from Cotgrove (1982) and Milbrath (1984). This was a single factor model assessing the respondents’ belief in the existence of specific environmental problems. The CFA indicated that the model fit very well with all fit statistics above .98. The RMSEA was slightly higher than desired at .09. All factor loadings were, however, above .62 with an average loading of .71. Coefficient alpha for the belief scale was .85 indicating high reliability. The environmental concern scale consisted of six items derived from Kilbourne, Beckmann, and Thelen (2002) and Milbrath (1984). The CFA for the scale indicated that 18 the model fit the data well with all fit statistics above .98 and an RMSEA of .045. Coefficient alpha for the concern scale was .81 and all factor loadings were above .61. The average loading was .63 for all items. Because measures of beliefs and concern are conceptually similar and sometimes used interchangeably (Fransson and Garling 1999), it was necessary to determine if they represented different measures. Exploratory factor analysis was used in which all twelve items were tested simultaneously. The results of that test indicated that there were two distinct factors with all belief items loading on one factor and all concern items loading on the second factor. The two factors together accounted for 54% of the variance. Thus, for this sample, beliefs are clearly distinguishable from concerns. The final construct assessed was the behavioral model consisting of two parts, indirect behavior and direct behavior. Each of these constructs contained three items. Because they are considered as distinct types of behavior, they were assessed separately. For direct behaviors, the three-item construct had fit statistics above .98 and an RMSEA near zero indicating good fit. To assess the model, the measurement error for two items shown to be equivalent was constrained to be equal to get positive degrees of freedom for the model. The same procedure was used for direct behavior that also contained three indicator items. The results were similar with fit statistics above .98 and an RMSEA of .078. Because both indirect and direct behaviors are measures of behavior, they formed a second order factor. The second order factor fit the data well with fit statistics above .98 and an RMSEA of .038, so this was the model used in the structural equation analysis. The results of the analysis of the measurement models indicated that all the constructs in the proposed model fit the data well as the individual constructs were within 19 the accepted ranges on all fit statistics. In the final model, the values construct was a second order factor with ST, SE, and OP as first order constructs. The NEP was a second order construct with balance of nature, limits to growth, and human domination as first order factors. Beliefs and concern were both first order constructs and behavior was a second order construct with direct and indirect environmental behaviors as first order factors. Structural Equation Model The proposed model, presented in Figure 2, was tested using AMOS 6.0 (Arbuckle 2005) structural equation modeling software. The theory developed suggests a hierarchical model beginning with general values that lead to more general beliefs and folk ecological wisdom. This leads to more specific beliefs about environmental problems that, in turn, create environmental concern in the individual. Environmental concern then results in two types of environmentally related behaviors, direct and indirect. Because of the scoring of the constructs, the hypotheses suggest that the links between the constructs should be both statistically significant and positive. The results of the test, included in Figure 1, confirm all of the hypotheses contained in the model. Overall, the structural model does fit the data acceptably. The fit statistics for the full model were CFI=.91, IFI=.91, and RMSEA=.042. All parameter estimates for indicator variables in the full model were statistically significant at less than .001. Further, all parameter estimates linking the constructs in the structural model were significant at less than .001 and were in the predicted directions. In all cases, the link 20 from a higher-level construct to a lower level construct was positive. As individuals have more self-transcendent and openness values, they are higher on the NEP scale. Those with higher NEP scores indicating belief in the balance of nature, limits to growth, and human domination are more likely to believe there are environmental problems. As belief in the existence of environmental problems increases, the individual becomes more concerned about the environment. As environmental concern increases, both direct and indirect self-reported environmentally related behaviors increase. This establishes the causal sequence among the constructs in the model. Numerous alternative models were tested that linked the constructs differently, but none were shown to achieve the statistical significance of the model proposed. Connecting non-adjacent constructs yielded no significant direct links. Discussion The results of this national study of environmental constructs provide support for the proposed hierarchical model in Figure 1. The model depicts a causal chain of environmentally related constructs beginning with values and ending with behaviors. At each level of the model, measures are more closely related to the environmental behaviors. The SVI is a second order factor model with three latent constructs, selftranscendence, self-enhancement, and openness to change. Each of these constructs relates to the more environmentally oriented constructs. This result is consistent with both Schultz and Zelezny (1999) and Stern and Dietz (1994) indicating that self- 21 transcendent values are positively related to the NEP that reflects general beliefs about the environment. In addition, the results indicated that openness values are related to the NEP as well. While this latter relationship has received inconsistent support in the past, it is not an unexpected finding having been shown by Kilbourne et al. (2001) to affect willingness to change behaviors related to the environment. Thus, when incorporated with self-transcendent and negatively scored self-enhancement measures, openness to change affects the NEP. The second-order NEP has three latent constructs referred to as balance of nature, human exemptionalism, and limits to growth. Each of the latent constructs contributes significantly to the NEP construct as a whole. This is consistent with previous use of the NEP in both its original and revised forms and parallels the work of Cordano, Welcomer, Scherer (2003). The causal link between environmentally related values and general ecological beliefs as measured by the NEP was positive (0.43) as hypothesized. This indicates that as one’s values related to the environment increase, their general environmental beliefs increase. Each of the first order constructs contributed significantly to the overall construct referred to as the NEP. For the next causal link, that between general ecological beliefs and one’s belief in the existence of specific environmental problems, the NEP positively affected specific beliefs (0.93) as predicted. The higher one’s belief in the NEP, the more they believed that specific environmental problems exist. This included such problems as ozone depletion, global warming, and species extinction. These environmental beliefs, it is argued in the model, are antecedent to environmental concern because to be concerned, one must believe that there is something 22 to be concerned about. Belief in the NEP and specific environmental problems, combined with high environmental values, would reasonable activate psychological mechanisms leading to increased concern. The belief that social and individual changes are necessary to ameliorate environmental decline reflects this environmental concern. Individuals who are concerned would be likely to indicate willingness to change their own behavior and suggest that larger social and political change would also be necessary. This indirect measure was positively related to belief in environmental problems (0.78). As belief in specific problems increases, one’s environmental concern increases as well. The final causal link in the proposed model hypothesized that increased levels of concern would result in an increase in self-reported direct and indirect environmentally related behaviors. The test confirmed the hypothesized positive relationship between concern and both direct and indirect behaviors. The parameter estimate for direct behavior was 0.37 and for indirect behavior was 0.38 indicating that as environmental concern increases, one’s environmentally related behavior also increases. Thus, all links in the causal model hypothesized were positive confirming the hierarchical model proposed at the outset. The result has both research and policy implications. From the standpoint of future research, it can be argued that the model originally proposed by Stern, Dietz, and Guagnano (1995) provides an appropriate framework from which to examine the human/environmental relationship. There is a hierarchical relation descending from individual values through general environmental beliefs, specific beliefs, environmental concern, and ending in environmentally related behaviors. This indicates that some approaches to research are too limited to achieve sufficiently robust results. Measuring 23 two levels of the model, for example, while revealing relationships would fail to reveal the role the other variables play in achieving that result. The causal modeling approach used here considers all the relationships simultaneously capturing the hierarchical nature of the relationship. Future research should consider both the nature of the models and the analytical methods used to assess them. Policy implications of the results are also relevant. The hierarchical nature of the model indicates that policy instruments should not be one dimensional or privilege one variable, such as technology, in any desired outcome (Taschian, Slama, and Taschian 1984). Policies aimed at increasing recycling, or some other environmentally related behavior, for example, are likely to be less effective than desired if they fail to consider the antecedent variables in the causal chain leading to behavior. The results of this study indicate that antecedent variables have effects on downstream variables that, if left unaddressed in policy instruments, may have attenuating effects on outcomes. For example, if one desires consumers to consume less or differently as an outcome of consumer policy, it would be necessary to address, not only consumers’ immediate behavior, but also their beliefs and values that precede that behavior. Admonishing consumers for consuming without initiating values intervention strategies would be less effective than a comprehensive approach addressing both behaviors and values. Americans have been demonstrated to be materialistic, and their materialism is related to self-enhancement and self-transcendence values (Kilbourne, Grünhagen, and Foley 2005). If materialism is a cultural characteristic that results in increased material consumption, environmental decline is the inevitable long-run result as indicated by Daly’s (1991) “impossibility theorem.” If policy instruments address the 24 lower levels of the model without addressing the higher levels, then the higher levels will consistently militate against the policy reducing its effectiveness. This is one possible explanation for the attitude-behavior gap evidenced in American society. While the hierarchical approach taken here represents an improvement on piecemeal approaches examining small parts of the environmental problem independently, it does not incorporate all possible levels of examination. Stern, Dietz, and Guagnano (1995), for example, argue that social-structural variables precede values in the hierarchical model. This level of the model was not examined and should be included in future research. It has also been argued by Kilbourne (1998) that the DSP has both cosmological and socio-demographic dimensions representing a larger domain that might be antecedent to and inform value development in individuals. Thus, self-enhancement of self-transcendent values, both important in environmental beliefs, concern, and behavior, do not arrive fully articulated in the individual. They are the product of centuries of cultural development in science, economics, and politics. This suggests that there may be a level in the model above values and social structural variables. This has received little examination to date. The model presented here is in the process of development and can be expanded and enhanced. A further limitation on the model presented is that it represents a uniquely American perspective. Whether the constructs used in the model are valid across cultures is a very important question as environmental problems are trans-boundary problems. Future research should examine environmental problems as cross-cultural problems. Because few of the constructs used in environmental research possess cross-cultural invariance, researchers face a significant problem here. This is the next critical step for 25 environmental researchers. As a final limitation, it must be pointed out that the statistical procedures used in the analysis do not provide evidence of causality, only correlation. Thus, it cannot be concluded that the structure tested links variables in a causal way. The results are, however, consistent with previous work suggesting causal relations. 26 References Ajzen, Icek and Martin A. Fishbein (1977), "Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research," Psychological Bulletin, 84, 888-918. Allen, Chris T., Roger J. Calantone, and Charles D. Schewe (1982), "Consumers' Attitudes about Energy Conservation in Sweden, Canada, and the United States, with Implications for Policymakers," Journal of Public Policy & Marketing, 1 (1), 57-67. Alwitt, Linda F. and Robert E. Pitts (1996), "Predicting Purchase Intentions for an Environmentally Sensitive Product," Journal of Consumer Psychology, 5 (1), 49-64. Arbuckle, James L. (2005), Amos 6.0 User's Guide. Chicago, IL: Amos Development Corporation. Bechtel, Robert B., Victor C. Verdugo, and Jose de Quiroz Pinheiro (1999), "Environmental Belief Systems: United States, Brazil, and Mexico," Journal of CrossCultural Psychology, 30 (1), 122-28. Berger, Ida E. and Ruth M. Corbin (1992), "Perceived consumer effectiveness and faith in others as moderators of environmentally responsible behaviors," Journal of Public Policy & Marketing, 11 (2), 79-89. 27 Brockett, Patrick L. and Linda L. Golden (1990), "How Public Policy Can Define the Marketplace: The Case of Pollution Liability Insurance in the 1980s," Journal of Public Policy & Marketing, 9 (1), 211-26. Byrne, Barbara M. (2001), Structural Equation Modeling With AMOS. Mahwah, NJ: Lawrence Erlbaum Associates. Cordano, Mark, Stephanie A. Welcomer, and Robert F. Scherer (2003), "An Analysis of the Predictive Validity of the New Ecological Paradigm Scale," Journal of Environmental Education, 34 (3), 22-28. Cotgrove, Stephen (1982), Catastrophe or Cornucopia: The Environment, Politics, and the Future. New York: Wiley. Daly, Herman E. (1991), Steady-State Economics. Washington, DC: Island Press. Dietz, Thomas and Paul C. Stern (1995), "Toward a Theory of Choice: Socially Embedded Preference Construction," Journal of Socio-Economics, 24 (2), 261-79. Dietz, Thomas, Paul C. Stern, and Gregory A. Guagnano (1998), "Social Structural and Social Psychological Bases of Environmental Concern," Environment & Behavior, 30 (4), 450-71. 28 Dobscha, Susan and Julie L. Ozanne (2001), "An Ecofeminist Analysis of Environmentally Sensitive Women Using Qualitative Methodology: The Emancipatory Potential of an Ecological Life," Journal of Public Policy & Marketing, 20 (2), 201-14. Dryzek, John S. (1987), Rational Ecology: Environment and Political Economy. Oxford: Oxford University Press. Dunlap, Riley E. and Kent D. Van Liere (1978), "The "New Environmental Paradigm": A Proposed Instrument and Preliminary Results," Journal of Environmental Education, 9 (1), 10-19. Dunlap, Riley E., Kent D. Van Liere, Angela G. Mertig, and Robert Emmet (2000), "Measuring Endorsement of the New Ecological Paradigm: A Revised NEP Scale," Journal of Social Issues, 56 (3), 425-42. Eckersley, Robyn (1992), Environmentalism and Political Theory: Toward an Ecocentric Approach. Albany: State University of New York Press. Fransson, Niklas and Tommy Garling (1999), "Environmental Concern: Conceptual Definitions, Measurement Methods, and Research Findings," Journal of Environmental Psychology, 19 (4), 369-82. 29 Granzin, Kent L. and Janeen E. Olsen (1991), "Characterizing Participants in Activities Protecting the Environment," Journal of Public Policy & Marketing, 10 (Fall), 1-27. Hair, Joseph F., Rolph E. Anderson, Ronald L. Tatham, and William C. Black (1998), Multivariate Data Analysis (5 ed.). Upper Saddle River, NJ: Prentice Hall. Inglehart, R. (1981), "Post-materialism in an Environment of Insecurity," American Political Science Review, 75, 880-900. Johnson, Cassandra Y., J. M. Bowker, and H. Ken Cordell (2004), "Ethnic variation in Environmental belief and Behavior: An Examination of the New Ecological Paradigm in a Social Psychological Context," Environment and Behavior, 36 (2), 157-86. Kempton, Willett, James S. Boster, and Jennifer A. Hartley (1995), Environmental Values in American Culture. Cambridge, MA: MIT Press. Kilbourne, William E. (1995), "Green Advertising: Salvation or Oxymoron?," Journal of Advertising, 24 (2), 7-19. ---- (1998), "Green Marketing: A Theoretical Perspective," Journal of Marketing Management, 14 (6), 641-55. 30 Kilbourne, William E. and Suzanne C. Beckmann (1998), "Review and Critical Assessment of Research on Marketing and the Environment," Journal of Marketing Management, 14 (6), 513-32. Kilbourne, William E., Suzanne C. Beckmann, Alan Lewis, and Ynte van Dam (2001), "A Multi-National Examination of the Role of the Dominant Social Paradigm in Environmental Attitudes of University Students," Environment & Behavior, 33 (2), 20928. Kilbourne, William E., Suzanne C. Beckmann, and Eva Thelen (2002), "The Role of the Dominant Social Paradigm in Environmental Attitudes: A Multi-National Examination," Journal of Business Research, 55 (3), 193-204. Kilbourne, William E., Marko Grünhagen, and Janice Foley (2005), "A Cross-Cultural Examination of the Relationship Between Materialism and Individual Values," Journal of Economic Psychology, 26 (5), 624-41. Mathews, Freya (1991), The Ecological Self. London, UK: Routledge. McCarty, John A. and L. J. Shrum (2001), "The Influence of Individualism, Collectivism, and Locus of Control On Environmental Beliefs and Behavior," Journal of Public Policy & Marketing, 20 (1), 93-104. 31 Milbrath, Lester (1984), Environmentalists: Vanguards for a New Society. Albany, NY: University of New York Press. Milfont, Taciano L. and John Duckitt (2004), "The Structure of Environmental Attitudes: A First - and Second - Order Confirmatory Factor Analysis," Journal of Environmental Psychology, 24 (3), 289-303. Peter, J. Paul (1979), "Reliability: A Review of Psychometric Basics and Recent Marketing Practices," Journal of Marketing Research, 16 (1), 6-17. Richins, Marsha L. (2004), "The Material Values Scale: Measurement Properties and Development of a Short Form," Journal of Consumer Research, 31 (1), 209-19. Schultz, P. Wesley and Lynnette C. Zelezny (1998), "Values and Proenvironmental Behavior: A Five-Country Survey," Journal of Cross-Cultural Psychology, 29 (4), 54058. Schultz, Wesley P. (2001), "The Structure of Environmental Concern: Concern for Self, Other People, and the Biosphere," Journal of Environmental Psychology, 21 (4), 327-39. Schultz, Wesley P. and Lynnette Zelezny (1999), "Values as Predictors of Environmental Attitudes: Evidence for Consistency Across 14 Countries," Journal of Environmental Psychology, 19, 255-63. 32 Schwartz, Shalom H. (1994), "Are There Universal Aspects of the Structure and Contents of Human Values?," Journal of Social Issues, 50 (4), 19-45. ---- (1968), "Awareness of Consequences and the Influence of Moral Norms on Interpersonal Behavior," Sociometry, 31, 355-69. ---- (1992), "Universals in the Content and Structure of Values: Theoretical Advances and Empirical Tests in 20 Countries," in Advances in Experimental Social Psychology, M. P. Zanna, ed. Vol. 25. Orlando, FL: Academic Press, 1-65. Schwepker, Charles H. and T. Bettina Cornwell (1991), "An Examination of Ecologically Concerned Consumers and Their Intention to Purchase Ecologically Packaged Products," Journal of Public Policy & Marketing, 10 (2), 77-101. Steg, Linda and Inge Sievers (2000), "Cultural Theory and Individual Perceptions of Environmental Risks," Environment and Behavior, 32 (2), 250-69. Stern, Paul C. (1992), "Psychological Dimensions of Global Environmental Change," Annual Review of Psychology, 43, 269-302. ---- (2000), "Toward a Coherent Theory of Environmentally Significant Behavior," Journal of Social Issues, 56 (3), 407-24. 33 Stern, Paul C. and Thomas Dietz (1994), "The Value Basis of Environmental Concern," Journal of Social Issues, 56, 1-20. Stern, Paul C., Thomas Dietz, and Gregory A. Guagnano (1998), "A Brief Inventory of Values," Educational & Psychological Measurement, 58 (6), 984-1001. ---- (1995), "The New Ecological Paradigm in Social-Psychological Context," Environment & Behavior, 27 (6), 723-43. Taschian, Armen, Mark E. Slama, and Roobian O. Taschian (1984), "Measuring Attitudes Toward Energy Conservation: Cynicism, Belief in Material Growth, and Faith in Technology," Journal of Public Policy & Marketing, 3 (1), 134-48. White, Lynn (1967), "The Historical Roots of Our Ecological Crisis," Science, 155, 1203-07. 34 Figure 1 Structural Equation Model peace justice balance transcend balance environment change .97 .91 creative varied modify openness curious .82 values .44 .39 nep exempt rule use .65 .56 authority influence interfere spaceship enhance limits industry .90 success pollution extinction chemical beliefs rainforest warming .74 ozone concern abuse consume concern political .75 social enforce behavior .49 .45 member contribute suscribe product indirect limits direct waste recycle 35 Figure 2 Schwartz’s Value Inventory SELF-TRANSCENDENCE BENevolence TRAdition Self-DIrection CONformity STImulation SECurity HEDonism ACHievement POWer SELF-ENHANCEMENT CONSERVATION OPENNESS TO CHANGE UNIversalism 36 Appendix Questionnaire Items Schwartz’s Value Inventory CFI= .97 IFI= .97 RMSEA=.044 Self-transcendence α=.71 A world at peace (free of war and conflict) Social justice (correcting injustice, care for the weak) Protecting the environment (preserving nature) Openness α=.56 Creativity (uniqueness, imagination) A varied life (filled with challenge, novelty and change) Curious (interested in everything, exploring) Self-enhancement α=.63 Authority (the right to lead or command) Influential (having an impact on people and events) Successful (achieving goals) New Environmental Paradigm CFI= .97 IFI=.97 RMSEA= .044 Balance of nature α=.77 The balance of nature is very delicate and easily upset When humans interfere with nature it often produces disastrous consequences If things continue on their present course, environmental problems will increase Human domination α=.80 Humans have the right to modify the natural environment to suit their needs Humans were meant to rule over the rest of nature Plants and animals exist primarily to be used by humans Limits to growth α=.61 There are limits to economic growth that we cannot go beyond The earth is like a spaceship with only limited room and resources Industrial growth must be limited so we will always have resources to use Beliefs CFI=.97 IFI=.97 RMSEA= .095 α=.85 Many types of pollution are rising to dangerous levels Some living things are being threatened with extinction Continued use of chemicals in business will damage the environment Destruction of rainforests will have negative environmental consequences Global warming is becoming a problem Ozone depletion is an important environmental problem Concern CFI= .99 IFI=.98 RMSEA=.047 I am very concerned about the environment Humans are severely abusing the environment Loading .63 .68 .70 .62 .51 .52 .58 .70 .52 .66 .79 .76 .63 .94 .72 .52 .59 .63 .74 .74 .68 .63 .66 .69 α=.81 .60 .64 37 I would be willing to reduce my consumption to help protect the environment Major political change is necessary to protect the natural environment Major social changes are necessary to protect the natural environment Environmental laws should be enforced more strongly Behavior CFI=.99 IFI=.98 RMSEA= .038 Direct α=.72 I buy environmentally friendly products when possible I reduce household waste I use products made from recycled material whenever possible Indirect α=.74 I am a member of an environmental organization I contribute money to an environmental organization I subscribe to an environmental magazine .67 .60 .69 .61 .69 .61 .73 .77 .71 .62