,řJ"- ";,--'•- '>,' - J ■ - '■ '- - • Žací -4»ä'"í"-íf-í '^ *£-'-'... .-«•-■ ' *■ t^j _ (s. ^iLř vTSM,.Sl.f««*H.lL,3L ^tlSi "".(5 ' JODY M. HINES, HAROLD R. HUNGERFORD, and AUDREY N. TOMERA ABSTRACT: Despite the wealth of information which exists concerning environmental behavior, it is not known which variable or variables appear to be most influential in motivating individuals to take responsible environmental action. A metaanalysis of environmental behavior research was undertaken in an attempt to determine this. An exhaustive search of the empirically based environmental behavior research conducted over the past decade yielded a substantial number of studies representative of a broad academic base. The characteristics and findings of these studies served as the data for the meta-analysis. As a result of the meta-analysis, the following variables were found to be associated with responsible environmental behavior: knowledge of issues, knowledge of action strategies, locus of control, attitudes, verbal commitment, and an individual's sense of responsibility. A model of predictors of environmental behavior is proposed. kver the past decade, support has steadily 'grown among environmental educators for the importance of developing individuals who behave responsibly toward the environment (Stapp 1969; Hendee 1972; Childress and Wert, 1976). This support has Jody M. Hines is an assistant professor in the Department of Teaching, Price Laboratory School, University of Northern Iowa. Harold R. Hungerford is a professor in the Department of Curriculum, Instruction, and Media, Southern Illinois University. Audrey N. Tomera was a professor in the Department of Curriculum, Instruction, and Media, Southern Illinois University, until her death in December, 1986. grown to such an extent that it can now be said that the development of environmentally responsible and active citizens has become the ultimate goal of environmental education (Hungerford and Peyton, 1976; Roth 1970; Stapp 1971). Despite agreement as to the importance of this goal, it has not, as yet, been achieved (Roth 1981; Hungerford and Volk, 1983). Curricular and instructional strategies which effectively lead to the development of environmentally responsible individuals have not been'implemented in our school systems. One of the major impediments to the accomplishment of this goal stems from a lack of knowledge of those factors which have formative effects on the development of environmentally responsible behavior (Linke 1980). This lack of knowledge does not, however, appear to be the result of a scarcity of research on the topic. Interest in environmental behavior research has JOURNAL OF ENVIRONMENTAL EDUCATION recently expanded to involve individuals in a wide array of academic fields, e.g., education, psychology, sociology, engineering, political science, business, forestry, and communications (Borden 1977). Consequently, the volume of research devoted to environmentally responsible behavior has burgeoned over the past ten years. However, the lack of substantive ties between many of these fields has led to a situation in which it is particularly difficult to remain informed about current developments in environmental behavior research. One may speculate that this lack of communication among researchers in different academic areas may partially account for the present status of research with regard to environmental behavior. That is, that while a tremendous variety of variables has been investigated in relation to behavior in an environmental context, there is at present no agreement among researchers as to which of these variables appear to be most strongly associated with responsible environmental behavior. Such information is vital to environmental educators for its potential in terms of providing a sound empirical base on which to construct appropriate curricula for the development of environmentally responsible and active individuals. The present study attempted to address this concern. The major goals of this study were to analyze and synthesize the environmental behavior research which had been reported since 1971 in an effort (1) to identify those variables which the research indicated were most strongly associated with responsible environmental behavior, (2) to determine the relative strengths of the relationships between each of these variables and environmental behavior, and (3) to formulate a model of environmental behavior representative of the findings synthesized in this research. The primary methodology employed in accomplishing these goals involved the use of the Schmidt-Hunter meta-analysis techniques (Hunter, Schmidt, and Jackson 1982). Meta-analysis of Research Meta-analysis is the term applied to groups of precise statistical methods designed to integrate empirical findings of studies addressing the same relationship. The explicit, unambiguous, and operationally defined methods associated with this approach to research integration make it far superior to the more subjective narrative discursive reviews of the literature which have traditionally been employed (Glass, McGaw, and Smith 1982). In the case of this study, this methodology allows the identification of those variables which have been found by the research to be associated with responsible environmental behavior. The methods also permit the determination of the relative strengths of the associations between each of these variables and responsible environmental behavior. While the characteristics of this methodology and the guidelines for employing it are well documented (see for example, Hunter et al. 1982 and Glass et al. 1982), an attempt will be made in this limited space to briefly acquaint the reader with the general concepts involved. Meta-analysis begins with the set of all studies that an investigator locates which provide empirical evidence bearing on the relationship of interest. Key characteristics and findings from each study are systematically recorded. Study findings are then converted to a common statistic, a point-biserial correlation coefficient in this case. Each statistic is examined across studies and its weighted mean and variance calculated. These values are then corrected for errors due to sampling and for errors due to differences in the reliabilities of the instruments. The resulting mean correlations and accompanying standard deviations are then examined and interpreted. An important concept associated with meta-analysis is that of the moderator variable. A moderator variable is considered to be any variable, other than those directly under study, which is found to impinge upon the relationship of interest. Moderator variables may consist of methodological differences in studies, differences in demographic composition of the individuals comprising the study samples, specificity of measures, or other factors which operate by attenuating the effects of the variables under study. Meta-analysis provides a method for establishing the relevance of potential moderator variables. The suspected moderator variable is used to split the studies into subsets which are then meta-analyzed separately. The operation of the particular moderator variable is confirmed in situations in which there are large differences in subset means. For example, in this study one suspected moderator variable was the use of self-reported measures of behavior as opposed to reliance upon actual behavior as an indicator of responsible environmental behavior. This suspicion was based on skepticism expressed by researchers regarding the validity of the use of self-reported behavior as an accurate indicator of actual behavior. It has been shown that what people indicate on a questionnaire is often inconsistent with their actual behaviors (e.g., Brickman 1972; Deutscher 1973; Wicker 1969, 1971). On this basis, the self-reported versus actual behavior discrepancy was tested to determine whether the use of these different types of behavior measures was indeed acting as a moderator variable. It was also suspected that the composition of the sample may have had moderating effects on the relationship between a number of variables and environmental behavior. In particular, it was suspected that different correlations would be obtained from studies which relied upon samples comprised either entirely, or in part, of individuals with known ties to environmental organizations (e.g., Sierra Club) than would be obtained from HÍNES, HUNGERFORD AND TOMERA 3 samples derived from the general population. Metaanalysis of the data was able to provide information concerning the operation of this suspected moderator variable. Methodology The steps followed in the analysis and synthesis of environmental behavior research were: (1) location of appropriate studies, (2) extraction of pertinent information from these studies, and (3) analysis and synthesis of the extracted information via the Schmidt-Hunter metaanalysis procedures. The location of studies involved an exhaustive search of the literature entailing the use of over fifteen secondary sources of information (e.g., Psychological Abstracts, GEO Abstracts, Educational Resources Information Center, Dissertations in Eco-Education, Ecology and the Environment: A Dissertation Bibliography). These sources cited published works, dissertations, government documents, and other fugitive literature. Reference lists for those studies located in this initial search of secondary sources of information were also reviewed in an effort to find additional environmental behavior studies which had not surfaced in the preliminary search of secondary sources. A list of 380 studies for possible inclusion in the data set emerged. Sixty-five of these works could not be located and were thus eliminated from consideration for inclusion in the data set. Each of the remaining 315 studies were read. Only those studies which assessed variables in association with responsible environmental behavior and which reported empirical data on this relationship were included in the data set. In addition, the search was restricted to studies which had been reported since 1971. These restrictions resulted in the elimination of 187 studies. The remaining 128 studies comprised the data for this study. Ninety-eight of these studies were journal articles, thirteen were reported as dissertations or theses, twelve appeared as unpublished manuscripts, and five of the studies were published in books. Information was extracted from these studies by recording pertinent characteristics and findings for each study on coding sheets. An analysis of these data resulted in the emergence of a number of major categories of variables which had been investigated in association with responsible environmental behavior. These categories were: (1) cognitive variables, (2) psycho-social variables, (3) demographic variables, and (4) a category of experimental studies comprised of behavioral intervention approaches and classroom strategies aimed at encouraging responsible environmental behavior. These categories of variables were further broken down into subcategories, each of which was meta-analyzed separately. In the end, fifteen separate variables were meta-analyzed in an effort to determine the strength of their associations with environmental behavior. The findings of the meta-analyses of the first three of these four categories of variables are presented and discussed below (see Table 1). TABLE 1. Summary of Meia-analysis Findings for Descriptive Studies Corrected Corrected Number correlation standard of values Variable coefficient deviation based ona Verbal commitment .491 .130 6 Locus of control .365 .121 14 Attitude .347 .224 51 Personal responsibility .328 .121 6 Knowledge .299 .195 17 Educational level .185 .122 11 Income .162 .084 10 Economic orientation .160 .118 6 Age -.151 .200 10 Gender .075 .084 4 a Several studies reported data on more than one variable. t í i: Meta-anaiysis of Cognitive Variables I Cognitive variables, for the purpose of this study, in- Í; eluded those factors pertaining to knowledge of the en- 't vironment or to some aspect of an environmental issue. ( This entailed not only knowledge of environmental problems and their consequences, but may also have pertained to knowledge of how to take action on a par- | ticular environmental problem. Seventeen study find- 1 ings which dealt with cognitive variables measured in 1 association with environmental behavior were coded. | All of these outcome measures were the result of t descriptive as opposed to experimental studies. j; Meta-analysis of these seventeen study findings Ě resulted in a corrected correlation coefficient of .299 í (SD = .195) between knowledge and environmental y, behavior. The positive correlation coefficient indicates |' that those individuals with greater knowledge of en- p vironmental issues and/or knowledge of how to take ac- \. tion on those issues were more likely to have reported t- engaging in responsible environmental behaviors than ľ were those who did not possess this knowledge. f" Further, meta-analysis of the cognitive data revealed í that the mode of behavior assessment was operating as a f; moderator variable in the knowledge-behavior relation- f ship. A higher corrected correlation coefficient (r = í .37, SD = .202) was obtained from analysis of those i studies which employed actual measures of behavior £ than was obtained from analysis of those studies which [i relied upon self-reported behavior assessments (r = f, .291, SD = .192). i 4 JOURNAL OF ENVIRONMENTAL EDUCATION The population sampled was also confirmed to have attenuated the knowledge-behavior relationship. A substantially larger average correlation coefficient was obtained from those studies whose samples were comprised either entirely, or in part, of individuals with known ties to environmental organizations (r = .691, SD = .063) than were obtained from samples derived from the general adult population (r = .268, SD = .173) or from samples comprised of children (r = .192, SD = .071). Meta-analysis of Psycho-social Variables Psycho-social variables included those factors related to individual personality characteristics, including the perceptions individuals have of themselves and of others. The psycho-social variables which were rneta-analyzed in relation to environmental behavior included attitudes, locus of control, economic orientation, personal responsibility, and verbal commitment. Because meta-analysis can only be performed on finding bearing on the same relationship, it was necessary to maintain these separate categories for the analysis of the psychosocial variables. A ttitude-behavior Relationship Attitudinal variables, for the purposes of this study, included those factors which dealt with the individual's feelings, pro or con, favorable or unfavorable, with regard to particular aspects of the environment or objects related to the environment. This category included assessments of general attitude toward the environment or toward ecology, as well as more specific attitudes such as attitudes toward the energy crisis, attitudes toward unleaded gasoline, and attitudes toward taking environmental action. No distinction was made between affective and cognitive components of attitudes. Fifty-one outcome measures on the attitude-behavior relationship were coded. Meta-analysis of the full set of these studies (see Table 1) resulted in a corrected correlation coefficient of .347 (SD = .224). These results indicated the existence of a relationship between attitude and behavior, in that those individuals with more positive attitudes were more likely to have reported engaging in responsible environmental behaviors than were individuals with less positive attitudes. Further examination of these data was conducted in an effort to determine the nature of the attitudes under study. It was found that there were essentially two types of attitudes studied by researchers: attitudes toward ecology and the environment as a whole, and attitudes toward taking environmental action (e.g., attitudes toward recycling, toward petitioning, toward conserving energy). Forty-two of the attitude studies which were coded dealt with attitudes toward the environment, while nine studies were concerned with attitudes toward taking action. A slightly stronger relationship was detected between attitude toward action and environmental behavior (r = .377, SD = .145) than was observed between attitude toward the environment in general and environmental behavior (r = .338, SD = .243). Thus, the research indicated that both of these types of attitudes were related to behavior in an environmental context. Several moderator variables were detected in the attitude-behavior relationship. As was the case with the knowledge-behavior studies, mode of behavior assessment was found to have attenuated the attitude-behavior relationship. Higher attitude-behavior correlations were obtained in situations in which actual behaviors were assessed (r = .427, SD = .290) than were obtained in instances in which behaviors were determined via self-reported means (r = .334, SD = .209). The population which was sampled was also found to have affected the strength of the relationship between attitude and environmental behavior. Meta-analysis of those studies which sampled populations comprised of individuals with ties to environmental organizations resulted in a higher correlation (r = .593, SD = .273) than was obtained from the analysis of studies which either sampled the general adult population (r = .328, SD = .202) or which sampled children (r = .328, SD = .198). As was the case with the knowledge-behavior relationship, the size of the corrected correlation coefficient for the studies which employed environmental group members was substantially higher than were those values obtained for the other two groups sampled. This indicated that the use of environmental group members did attenuate the relationship between attitudes and environmental behavior. Locus of Control-behavior Relationship Locus of control is a general concept which is not restricted to behavior in an environmental context. Locus of control represents an individual's perception of whether or not he or she has the ability to bring about change through his or her own behavior. The concept is based on the belief that some individuals do not attempt to bring about change because they attribute change to chance or to powerful others (e.g., God, parents, government) rather than to their own behaviors. This perception is termed an external locus of control. Internal locus of control individuals, on the other hand, believe that their activities are likely to have an impact (Peyton and Miller, 1980). Six studies were coded which dealt specifically with the relationship between individual locus of control and responsibile environmental behavior. An additional nine studies were located which assessed what was referred to as an individual's "efficacy perception." This variable was defined as an individual's perception of his or her effectiveness in a given situation. Despite the dif- . rv\r.: „.:?-:vl:tf-&-:fti--vr&íř <&.*-m'& .;•■*•■ - '(- ■.--, ,.< «« ■■ . ■ ^;. HINES, HUNGERFORD AND TOMERA Situational factors Action skills - Knowledge of action strategies Knowledge of issues '' Attitudes * Intention to act Responsible environmental behavior Locus of control Personality factors Personal responsibility FIGURE 1. The Proposed Model of Responsible Environmental Behavior analyzed in this study. The erroneous assumption is often made that skills evolve naturally from knowledge. However, the Ramsey (1979) and Klingler (1980) studies provide strong evidence that this is likely not the case. In these studies it was found that treatments employing both knowledge and cognitive skill components resulted in significant increases in the numbers of actions reported as compared to those programs which employed only cognitive knowledge components. Thus it appears that skill in the application of action strategies to issues, combined with the appropriate knowledge, endow individuals with the abilities to take action. Abilities alone are not sufficient to lead to action. In addition, an individual must possess a desire to act. One's desire to act appears to be affected by a host of personality factors. These include locus of control, attitudes, and personal responsibility. Thus, an individual with an internal locus of control, positive attitudes toward the environment and toward taking action, and with a sense of obligation toward the environment will likely develop a desire to take action. If the requisite abilities to act are also present, action will likely follow. If these abilities are absent, action will not be likely to follow. The pathway discussed thus far indicates those factors which appear to be most strongly associated with responsible environmental behavior and their suspected mode of operation. One remaining category exists which can interupt this pathway to action—these are situational factors. Situational factors, such as economic constraints, social pressures and opportunities to choose different actions, may enter the picture and serve to either counteract or to strengthen the variables in the model. For example, if an individual has the cognitive ability, desire, and opportunity to help stop pollution by contributing to a local toxic waste fund, but simply cannot afford to do so, that person will not engage in the environmental action and, in this instance, the model's main pathway will not be followed. Situational factors may not only decrease, but may also increase the incidence of responsible environmental behaviors. For instance, a person may curb energy consumption only to save money and to collect the incentives offered in association with reduced consumption. While this person obviously posseses the knowledge and abilities to conserve, his actions have likely not stemmed from a deep-seated desire to conserve fossil fuels, but rather from personal and financial bases. Thus, in situations in which individuals do not possess those personality characteristics which would lead to the development of a desire to help alleviate environmental problems, these individuals may be enticed into behaving responsibly toward the environment by the application of behavioral intervention strategies. The complexity of the environmental behavior model and the operation of ever-changing situational factors 8 JOURNAL OF ENVIRONMENTAL EDUCATION illustrate the uncertainty involved in the prediction of environmental behavior. However, the model indicates several areas which are amenable to change by the efforts of environmental educators. The knowledge and skill components, and perhaps the personality components of the model, may be affected through the efforts of environmental educators. Approaches which address both affective and cognitive experiences and which provide individuals with opportunities to develop and to practice those skills necessary to lead to environmental action must be developed and implemented in our school systems. The knowledge and skill components can be addressed via issue identification, issue investigation, and action-taking approaches. The personality components of the model, however, are not as readily influenced through educational efforts. Definitive research concerning those factors which lead to the development of an internal locus of control, positive attitudes toward the environment and toward action-taking, and a sense of responsibility are needed to direct the efforts of environmental educators. It is not known at what point a person will forego economic and other personal benefits to do what preserves the integrity and stability of the environment. While the pathway represented in the model by knowledge, skills, and personality factors is the more desirable pathway by which to encourage environmentally responsible behavior, it may be more efficacious, in the case of certain environmental problems, to manipulate situational factors in order to produce the desired behavior changes. It has long been known that the prediction of behavior is an extremely complex process which is based on a multitude of factors. The findings of this meta-analysis and the subsequent development of an environmental behavior model serve to narrow the focus of the environmental behavior picture by determining those factors which appear to be most strongly associated with responsible environmental behavior. However, additional research is needed in an effort to discover those interrelationships which exist between each of the variables in the model. To accomplish this, research efforts must concentrate on all factors in the environmental behavior picture, rather than continuing to isolate individual components from those variables with which they likely interact. REFERENCES Asch, J. and B. M. Shore. 1975. Conservation behavior as the outcome of environmental education. Journal of Environmental Education, 6(4):25-33. Bickman, L. 1972. Environmental attitudes and actions. The Journal of Social Psychology, S7(2):323-24. Borden, R. J. 1977. One more look at social and environmental psychology: Away from the looking glass and into the future. Personality and Social Psychology Bulletin, 3:407-11. Childress, R. B., and J. Wert. 1976. Challenges for environmental education planners. Journal of Environmental Education, 7(4): 2-6. Duetscher, I. 1973. What we say/what we do: Sentiments and acts. Glenview, IL: Scott, Foresman, and Co. Glass, G. V., B. McGaw, and M. L. Smith. 1982. Meta-analysis in social research, Beverly Hills, CA: Sage. Heberlein, T. A. and J. S. Black. 1976. Attitudinal specificity and the prediction of behavior in a field setting. Journal of Personality and Social Psychology, 33(4):474-79. Hendee, J. C. 1972. Challenging the folklore of environmental education. The Journal of Environmental Education, 3:19-23. Hines, J. M. 1984. An analysis and synthesis of research on responsible environmental behavior. Ph.D. diss., Southern Illinois University at Carbondale. Hungerford, H. R. and R. B. Peyton. 1976. Teaching environmental education. Portland, ME: J. Weston Walch. Hungerford, H. R., and G. L. Volk. 1983. The challenges of K-12 environmental education. Paper presented for a National Association of Environmental Education Monograph. Hunter, J. E., F. L. Schmidt and G. B. Jackson. 1982. Metaanalysis: Cumulating research findings across studies. Beverly Hills, CA: Sage. Klingler, G. 1980. The effect of an instructional sequence on the environmental action skills of a sample of Southern Illinois eighth graders. Unpublished master's, research, Southern Illinois University at Carbondale. Linke, R. D. 1981. Achievements and aspirations in Australia EE. Journal of Environmental Education, 12(2):20-23. McCutcheon, L. F. 1981. Influences of energy conservtion education on attitudes and behaviors of selected youths in Piedmont, North Carolina. (Ph.D. diss., University of North Carolina at Greensboro. Dissertation Abstracts International, 1981, 42:\5\5A. Partain, J. D. 1979. An assessment of energy cognizance, attitude toward energy conservation, and perceived energy conservation behavior of selected public school educators. Ph.D. diss., Texas A&M University, 1979. Peyton, R. B., and B. A. Miller. 1980. Developing an internal locus of control as a prerequisite to environmental action taking. In Current Issues VI: The Yearbook of Environmental Education and Environmental Studies, edited by A. B. Sacks et al., 173-192. Columbus, Ohio: Clearinghouse for Science, Mathematics, and Environmental Education. Ramsey, J. M. 1979. A comparison of the effects qf environmental action instruction and environmental case study instruction on the overt behavior of eighth grade students. Unpublished master's thesis, Southern Illinois University at Carbondale. Roth, R. E. 1970. Fundamental concepts for environmental management education (K-16). The Journal of Environmental Education, 7:65-74. Roth, R. E. 1981. The whole earth: An EE perspective. Journal of Environmental Education, 12(2): 1-2. Stapp, W. B. Environmental encounters. In Outlines of Environmental Education, edited by C. Schoenfeld. Madison, WI: Dembar Educational Research Services, Inc. Stapp, W. B. et al. 1969. The concept of environmental education. Journal of Environmental Education, 7(1):30-31. Van Liere, K. D., and R. E. Dunlap. 1981. Environmental concern: Does it make a difference? How it is measured? Environment and Behavior, 13(6): 651-676. Wicker, A. W. 1969. Attitudes versus actions: The relationship of verbal and overt behavioral responses to attitude objects. Journal of Social Issues, 25:41-78. Wicker, A. W. 1971. An examination of the "other variables" explanation of attitude behavior inconsistency. Journal of Personality and Social Psychology, 7.9(l):18-30.