CORRELATES OF ENVIRONMENTAL BEHAVIORS Bringing Back Social Context

Surveys are an efficient and convenient means of collecting data on individuals’ environmental concern and environmental behavior, two domains between which a tenuous relationship is often observed. One aspect of tenuousness is addressed by identifying distinct subdimensions of self-reported private environmental behaviors. Survey methods more often than not fail to account for the social context within which individuals are environmentally concerned and behave in an environmentally friendly way. The problem of social context is addressed by developing a measure of social networks that includes participation, volunteering, and face-to-face contact with friends in environmental organizations. Evidence is taken from surveys among organized environmentalists and the general population in Norway in 1995 (N = 3,111). Social context is the only variable that significantly augments environmental behaviors across all subdimensions. Its effect is comparable to sociodemography, political attitudes, and environmental knowledge and concern combined. The field of environmental research has made some progress in identifying correlates of environmental concern (i.e., age, education and radicalism). But the literature is less consistent in identifying correlates of environmental behavior. It has also been shown that environmental concern sometimes fails to predict such behavior. 1 Studies have therefore concluded that the 181 ENVIRONMENT AND BEHAVIOR, Vol. 33 No. 3, March 2001 181-208 © 2001 Sage Publications, Inc. environmental attitude-behavior correspondence (ABC) is tenuous (e.g., Ajzen & Fishbein, 1977; Axelrod & Lehman, 1993; Diekmann & Preisendorfer, 1998; Kraus, 1995; Newhouse, 1990; Schultz & Oskamp, 1996). If a weak ABC really exists, researchers will unearth a tenuous relationship between attitudes and behaviors regardless of how it is measured. Tenuousness could be reduced, however, if measurement were more precise and detailed. If a strong ABC exists, but empirical results are weak, three possible explanations may apply. First, researchers might expect too much from statistical estimates. For example, they might underestimate empirical relationships through the correlation coefficient (Achen, 1977; Kraus, 1995; Ozer, 1985; Rosenthal, 1990). For example, Andrews and Withey (1974, pp. 13-16) report that two identical questions on life satisfaction posed at an interval of approximately 10 minutes in the same survey only correlated at .61. Although the correlation entails that 92% of the respondents ticked the same or an adjacent category on the 7-point scale, it nevertheless means that the item at Time 1 statistically explains 37% of the identical item at Time 2. In the context of surveys, therefore, a correlation coefficient of .61 represents an attitudinal tautology. Second, measurement might be technically incorrect and/or conceptually imprecise. The observed relationship between attitudes and behaviors becomes statistically and empirically stronger when it is correctly measured at the same level of specificity (Ajzen & Fishbein, 1977; Axelrod & Lehman, 1993; Kaiser, Wolfing, & Fuhrer, 1999; Kraus, 1995; Steel, 1996; Vining & Ebreo, 1992); for example, “attitude and intentions to recycle household newspapers were significant predictors of recycling behavior” (Boldero, 1995, p. 440). Third, social context and external factors might be neglected. Because a number of environmentally friendly activities are performed for rather different reasons within different contexts (Hallin, 1995; Mårtensson & Petterson, 1997), low ABCs may stem from a failure to account for social contexts and potentially influencing factors external to the relationship under study. The aim of this article is to contribute to the field of environmental behavior studies by exploring social context through Norwegian surveys on en182 ENVIRONMENT AND BEHAVIOR / March 2001 AUTHORS’ NOTE: Some of the analyses in this article were presented at Society, Environment, and Sustainability—The Nordic Perspective, Oslo, 25-27 August 1997. The authors thank Ian B. Couchman and one anonymous reviewer for useful comments on a previous version. Correspondence should be addressed to Gunnar Grendstad, Christiesgt.15, N-5007 Bergen, Norway, or e-mail at gunnar.grendstad@ isp.uib.no. vironmentalism. We examine both general and more specific private behaviors and analyze the extent to which they can be explained. We complement the correlates of environmental and ecological attitudes with sociodemographics, political attitudes, environmental knowledge, and social context. Social context is measured through participation in environmental organizations, frequency of volunteering, and face-to-face contact with friends in the organization. Our measure of social context such as participation allows us to distinguish the general public from active as well as inactive members of environmental organizations. CORRELATES OF ENVIRONMENTAL BEHAVIOR It is not implausible to assume that correlates of environmental concern (i.e., low age, political radicalism and high education, Jones & Dunlap, 1992) are also closely related to environmental behavior. But research has shown that the two sets of correlates are not always identical (e.g., Nord, Luloff, & Bridger, 1998). In a meta-analysis, Hines, Hungerford, and Tomera (1986/1987) showed that verbal commitment was most strongly correlated (i.e., corrected correlation coefficients) with environmental behavior (r = .49), followed by attitude (r = .35), knowledge (r = .30), educational level (r = .19), and income (r = .16). Age and gender were statistically unrelated to environmental behavior in their analyses. In another meta-analysis, where attitudes and behavior were found to correlate on average at .38, Kraus (1995, p. 69) concluded that ABC will increase when behavior is self-reported and when respondents are nonstudents. ABC has also been shown to improve with level of specificity and when behaviors are easily performed (Stern, 1992). In this article, we select and review four sets of correlates: sociodemographics, political attitudes, environmental attitudes (including environmental knowledge), and social context. Although some progress has been made in identifying the sociodemographic correlates of environmental concern (e.g., Jones & Dunlap, 1992; Lowe & Rüdig, 1986; Milbrath, 1984; Stern & Dietz, 1994; Van Liere & Dunlap, 1980; Van Liere & Dunlap, 1981), there has been less empirical and theoretical progress in examining the relationship between these correlates and environmental behavior. The relationship between gender and environmental concern is uncertain (Davidson & Freudenburg, 1996; Mohai, 1992; Stern, Dietz, & Kalof, 1993). Ecofeminists have claimed that women are able to understand relationships in nature better than men by virtue of their role as life-givers and their Olli et al. / ENVIRONMENTAL BEHAVIORS 183