The Globalization of Environmental Concern and The Limits of The Postmaterialist Values Explanation: Evidence from Four Multinational Surveys

Abstract

Conventional wisdom has long held that widespread citizen concern for environmental quality is limited to wealthy nations. Both academics and policymakers assume that residents of poor nations are too preoccupied with satisfying their “material” needs to support the “postmaterialist” value of environmental protection. This view was challenged by results of Gallup's 24-nation “Health of the Planet” (HOP) survey conducted in 1992, as the HOP found highly inconsistent and often negative correlations between national affluence and environmental concern. The current article compares results from three waves of the “World Values Survey” (WVS) to those of the HOP. When appropriate measures of environmental concern are employed, the WVS results generally replicate those of the HOP, as in all three waves such concern correlates inconsistently with national affluence. The overall results suggest that citizen concern for the environment is not dependent on national affluence, nor on affluence-based postmaterialist values.

ACKNOWLEDGMENTS

Thanks are extended to the Finnish Social Science Data Archive at the University of Tampere for making the World Values Survey data available, and to Toralf Zschau and anonymous reviewers for helpful comments on an earlier draft.

NOTES

Notes

¹ Like most advocates of conventional wisdom, Beckerman implicitly conflates the views of citizens of poor nations with those of their governmental leaders. Yet, as Broad and Cavanagh (1993:66) emphasize, the extreme degree of inequality in many poor and developing nations often yields governments that serve the interests of economic elites by pursuing environmentally harmful policies aimed at maximizing exploitation of natural resources that are not necessarily favored by the general public. We focus on citizens' views, both because they are an essential component of conventional wisdom and on the assumption that even in nondemocratic nations citizen preferences may have some impact on policymaking.

² This is not to deny that there were major divisions between rich and poor nations in Rio, but their differences were more over diagnoses of the causes of environmental problems and policies and funding for environmental protection rather than over the seriousness of environmental problems (see, e.g., Haas, Levy, and Parsons 1992). Also, it is important to note that survey data reveal only modest differences between the general publics of rich and poor nations on such issues (Dunlap 1994).

³ In recent years questions have been raised about the utility of the postmaterialist values explanation of environmental concern (Dietz et al. 2005), as some studies have found very weak correlations between endorsement of postmaterial values and support for environmental protection among general publics (e.g., Kidd and Lee 1997; Dietz, Stern, and Guagnano 1998). It may be that postmaterial values were a better predictor of activism, such as membership in environmental organizations or green parties (Inglehart 1977, 1990), than they are of pro-environmental opinions among the general public.

⁴ In parallel fashion, Diekmann and Franzen (1999:541) drew on rational choice theory to argue that “Standard economic reasoning suggests that restoration of a damaged environment is not only a collective good but also a ‘superior' good, that is, demand rises with income.” This assumption led them to question the HOP results and to provide results from 24 nations surveyed in the 1993 International Social Survey Project (ISSP) that purportedly demonstrate a positive correlation between national affluence and citizen concern for the environment. However, we have shown that the measure they employ is inherently flawed (clearly multidimensional rather than unidimensional) and that when more appropriate indicators of environmental concern are constructed from the ISSP data, the relationships between national affluence and environmental concern are inconsistent and not all that different from those of the HOP and the three waves of the WVS reported in this article (Dunlap and York 2008). In contrast to Diekmann and Franzen's (1999) claims, Kemmelmeier et al.'s (2002) finding that a narrow measure tapping willingness to make economic sacrifices for environmental protection constructed from the ISSP data is positively correlated with national affluence appears legitimate and compatible with the HOP findings.

⁵ Inglehart (1995:58) and others refer to this variously as the “1990–1991” and the “1990–1993” wave of the WVS, but we believe the latter is correct.

⁶ Inglehart did not refer to reports of the HOP results (e.g., Dunlap et al. 1993) nor to Brechin and Kempton's (1994) use of them, despite the emphasis given to objective conditions in both of these publications.

⁷ Because of the complexity of the construct, measures can vary dramatically in terms of substantive content (from broad coverage of “environmental quality” to a focus on “pollution” or “resource depletion,” or even a narrower focus on, e.g., “acid rain”), geographical scale (e.g., local, national, and global), and mode by which levels of “concern” are elicited—ranging from perceived seriousness to support for governmental regulations to personal actions. Further, the diverse conceptualizations and operationalizations of environmental concern appear to lead to inconsistencies in empirical findings regarding, for example, the social bases of such concern (see Dunlap and Jones 2002).

⁸ Another criterion for inclusion of nations in the analyses was that data on their per capita GDP is available for the relevant years of 1990, 1995, and 2000 (as reported in World Bank 2003). Also, some subnational regions such as the Basque region of Spain for which data were collected separately were deleted. Finally, the WVS sampled Northern Ireland and Britain separately and East and West Germany separately. We deleted Northern Ireland and East Germany, and assigned the UK's GDP per capita value to Britain and Germany's GDP per capita value to West Germany.

⁹ The survey also included two relevant “behavioral” items, as respondents were given a long list of “voluntary organizations and activities” and asked to indicate, first, “which, if any, do you belong to?” and, second, “which, if any, are you currently doing unpaid voluntary work for?” The list included “conservation, the environment, ecology” organizations. However, since the percentages indicating membership exceeded 5 percent in only a handful of nations, and the percentages indicating voluntary work were even lower, there is insufficient variance on these two items to create useful measures.

¹⁰ We found two factors in 33 nations and three factors in three (Belarus, Spain, and Portugal). In all of the nations with two factors, the “give income” and “pay taxes” items (see item labels in Table 1) loaded strongly on the “willingness to make economic sacrifices” factor (average loadings across nations for each item on this factor were greater than 0.83) and weakly on the other factor (average loadings of −0.06 and −0.08, respectively). Likewise, in these 33 nations “unemployment,” “too anxious,” and “not urgent” loaded strongly on the “perceived seriousness of environmental problems” factor (average loadings greater than 0.68) and weakly on the other factor (average loadings −0.12, −0.03, and 0.07, respectively).

The ambiguous “not cost” item loaded less consistently than the other items, loading more strongly on the willingness factor in 27 nations and more strongly on the seriousness factor in six nations. Furthermore, its loadings were not as clearly dichotomized as those of the other items, loading on average 0.58 on the willingness factor and 0.20 on the seriousness factor.

In the three nations where three factors emerged, “give income” and “pay taxes” formed a factor as in the other 33 nations, “unemployment” and “not urgent” formed a factor similar to the second factor in those 33 nations, while “not cost” and “too anxious” formed a third factor—although “too anxious” loaded moderately on all three factors. Thus, deleting the “not cost” item and creating the same two factors in these three nations as in the other 33 nations does not do violence to the data.

¹¹ The distinct nature of the two indexes we created, the Willingness Index and the Seriousness Index, is also affirmed by correlations among individual respondents. When we pool all individuals across nations (N > 40,000), the correlation between the two indexes is 0.04, while the correlations between “given income” and “pay taxes” is 0.58 and the correlations among “unemployment,” “too anxious” and “not urgent” all exceed 0.25—quite reasonable for inter-item correlations.

¹² The inadequacy of Inglehart's (1995) four-item index is reflected by the fact that at the national level it correlates +0.73 with the Willingness Index (with which it shares two items), but only +0.37 with the Seriousness Index and −0.20 with approval of the ecology movement.

¹³ Dunlap and Mertig (1995) used GNP per capita, but GDP is now seen as a better measure of national affluence and thus we use GDP per capita for the WVS analyses.

¹⁴ The analyses typically yielded three dimensions (as presented in Table 2) that are consistent with the substantive meaning of the items: “willingness to make economic sacrifices,” “green consumerism,” and “environmental activism.” This pattern was fairly consistent across nations, with the least stable items being “protect environment” and “reduce water use,” particularly the former. In 30 of these 36 nations three factors emerged. In 16 of these nations the items loaded most strongly on the “expected” dimensions presented in Table 2. In six other nations the expected pattern emerged, except “protect environment” loaded more strongly on the “green consumerism” factor than on the “willingness” factor. In four other nations the same pattern emerged, except “protect environment” loaded more strongly on the “environmental activism” factor. In the remaining four nations for which three factors emerged, the loadings were less clear, although in all of them “pay taxes” and “higher prices” consistently formed their own dimension.

The average factor loadings across nations (for the 30 with three factors) confirm the dimensions presented in Table 2, but also point to the instability of “protect environment.” “Pay taxes” and “higher prices” each had average loadings of greater than 0.85 on the “willingness” dimension, and average loadings between −0.03 and 0.01 on the other dimensions. “Protect environment” had average loadings of 0.37, 0.17, and −0.01 on the “willingness,” “green consumerism” and “environmental activism” dimensions, respectively, leading to our decision not to include it in the Willingness Index. “Household products,” “reuse/recycle,” and “reduce water use” each had average loadings on the “green consumerism” dimension of greater than 0.62, and between −0.03 and 0.14 on the other dimensions. The two activism items had average loadings of greater than 0.76 on the “environmental activism” dimension and between −0.01 and 0.03 on the other dimensions.

In one nation (Armenia) four factors emerged, with “pay taxes” and “higher prices” loading heavily on one factor, “protect environment” and “reduce water use” on a second, “household products” and “reuse/recycle” on a third, and “meeting/letter/petition” and “contributed to organization” on a fourth. In five nations (Bangladesh, Peru, Nigeria, Norway, and Japan) two factors emerged. In four of these nations the three willingness items formed one dimension and the other items formed another. In Norway the first two willingness items, “pay taxes” and “higher prices,” formed their own dimension and all other items loaded more strongly on the other dimension.

Overall we feel that creation of the three indexes reported in Table 2 (along with the individual ecology movement item) across all 36 nations is justifiable. Inter-item correlations from the pooled individual-level data set (N > 42,000) confirm the appropriateness of these indexes, as each item correlated more highly with the other items in its dimension than it did with items in other dimensions. The individual-level correlations among the three indexes ranged from 0.15 to 0.27.

¹⁵ This is mirrored by the fact that while aggregate, national-level responses to the first two items correlate very highly (0.74) across nations, the third (protect environment) item correlates more modestly with the first (0.48) and especially the second (0.31) item.

¹⁶ As a reviewer pointed out, the three green consumerism items are also double-barreled in that they ask respondents to report whether or not they engaged in behaviors and their motives for engaging in the behaviors. People who conserve for economic reasons (likely in poor nations) may have a difficult time deciding how to respond.

¹⁷ Since Puerto Rico was surveyed separately from the United States in both the second and third waves of the WVS, we included it here and in our analysis of the third wave data as a separate nation.

¹⁸ The Willingness Index correlates 0.22 (n.s.) with the Green Consumerism Index, 0.28 (p < 0.05) with the Environmental Activism Index, and 0.32 (p < 0.05) with Confidence in the Ecology Movement. The latter item correlates 0.22 (n.s.) with Green Consumerism and 0.14 (n.s.) with Environmental Activism. The correlation between the Green Consumerism and Environmental Activism Indexes is noticeably higher, 0.59 (p < .01), not surprising since both are formed from a list of self-reported behavioral items.

¹⁹ In this vein it should be noted that Brechin and Kempton (1994:259) provide insightful evidence from a limited multinational Harris survey showing that citizens of poor nations are significantly more willing than those in rich nations to express willingness to contribute time on behalf of environmental protection, whereas the reverse is true for willingness to pay higher taxes. This emphasizes the importance of measuring commitment to environmental protection in a nonmonetary fashion. Of course, in the WVS even the “willingness to pay” items do not produce the expected positive relationships between national affluence and citizen concern for the environment.

²⁰ Because these four items have been included in factor analyses of the two prior waves and because we wanted to create a Willingness Index in the third wave similar to those used in the first two waves, we did not factor analyze the items in the third wave. Our decision to combine “give income” and “pay taxes” into a Willingness Index is affirmed by the pattern of inter-item correlations among all four items in the pooled data set of all individuals across nations (N > 26,000). The correlation between give income and pay taxes is 0.63, whereas all other inter-item correlations are less than 0.30.