Environmentally Significant Consumption: Research Directions (1997)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

124 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION 5 Strategies for Setting Research Priorities Paul C. Stern, Thomas Dietz, Vernon W. Ruttan, Robert H. Socolow, and James L. Sweeney The topic “environmental impacts of consumption” suggests a great variety of possible research directions, a few of which are illustrated in Chapters 3 and 4. Because of this variety and the persistent confusion about definitions of consumption (see Chapter 2), it is useful to apply an importance criterion in considering priorities for research on consump- tion. The highest-priority research questions should be those that con- cern aspects of consumption with major environmental effects. This chapter discusses three strategies for setting research priorities so as to direct attention to the most important links between consumption and the environment—that is, to the factors that account for the kinds of consumption that have major deleterious effects on the environment or that may have such effects in the future. It focuses especially on a re- search strategy that first considers the major environmental effects of consumption and then reasons from environmental effects to human causes. It also discusses two other strategies: one that proceeds from human causes to their environmental effects, and one that begins with possible policy interventions and considers their effects. ENVIRONMENT-FIRST STRATEGY FOR SETTING RESEARCH PRIORITIES Human activities that directly perturb key properties of the biophysi- cal environment are important research topics under an importance crite- rion. Such activities have been referred to as proximate causes of environ- 124 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 125 Sources of Change Within and Between Environmental Systems Environmental Systems Human Proximate Proximate Effects on What Causes Humans Value Human Systems Sources of Change Within and Between Human Systems FIGURE 5-1 Interactions between human and environmental systems. SOURCE: National Research Council (1992). mental change (see Figure 5-1), to distinguish them from other activities, called driving forces, that have environmental effects only through their influence on proximate causes (National Research Council, 1992). Hu- man beings proximately cause environmental change when they burn fuel and thus release combustion products into the atmosphere; when they irrigate land with river water, changing the composition of the water and the height of the water table; when they cut trees or otherwise alter land cover and the habitats of biota; and in countless other ways. The immediate purposes for which they do these things (travel, crop produc- tion, construction, etc.) and the deeper reasons behind those purposes (subsistence, financial gain, self-expression, status seeking, etc.) cause environmental change only indirectly. This line of thinking, from environmental effects to proximate causes to driving forces, suggests a promising and systematic strategy for select- 

126 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION ing research that is likely to have relatively large practical importance. The strategy begins by identifying the proximate causes of an environ- mental condition of concern (the human activities that have the greatest direct effect on it) and then addresses, in a more or less logical order, the following questions: Which actors are responsible for these activities? What driving forces govern their behavior and how do these forces affect each other? What are the trends in these activities and their driving forces over time? How might the activities be changed, if change is desired? The strategy has been described in detail in previous work on the causes of global environmental change (National Research Council, 1992) and is summarized here as it applies to questions of consumption. Identifying Proximate Causes of Environmental Change A logical first step in employing this strategy is to identify important environmental properties that human activity may be changing. For ex- ample, for climate change, the key environmental properties are atmo- spheric concentrations of carbon dioxide and other greenhouse gases, and the albedo of the earth. Information on such properties comes from re- search in the relevant natural sciences, which helps set the agenda for research on human activity by directing attention to those human activi- ties that directly affect the important environmental systems. The proximate causes of changes in greenhouse gas concentrations are known to a good first approximation. The most important single one by far is the burning of fossil fuels which, by releasing carbon dioxide and methane, is currently responsible for about half of all anthropogenic greenhouse gas emissions. Other important proximate causes include cattle raising, certain land use changes, such as clearing forests and creat- ing or eliminating wetlands, and the various activities that release chlo- rofluorocarbons. (Some land use changes also affect global climate by altering the earth’s albedo.) The uncertainties in estimates of the impor- tance of each of these causes can be reduced by further research on the relevant biophysical processes. Nevertheless, the broad picture can be summarized well enough for the purposes of setting initial priorities for research on human activities. The summary is graphically represented in Figure 5-2, which presents the proximate causes as the limbs of a tree, with their thicknesses indicating their relative contributions to climate change induced by greenhouse gas emissions. Greenhouse gas emissions provide a good illustration of the strategy of reasoning backward from environmental conditions to their causes. For many other major environmental changes, the relative impor- tance of various human activities as proximate causes is not as well un- derstood. For example, the proximate causes of species extinctions in- 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 127 Process Steam Motors Other Heat Power Refrigeration Aerosols Industrial Uses Appliances Industry Cooling .20 Heating Paddy Rice Cattle Nitrogen Buildings .03 .03 Fertilization Other .02 .14 Land Biomass Clearing Fuel Air Rail Transportation Landfills .01 .12 Truck Autos Other .05 Agriculture .08 Landfills Fossil Fuel CFCs Biomass Other Use .25 Burning .06 .46 .15 FIGURE 5-2 A representation of the relative contributions of human activities in the late 1980s to global emissions of greenhouse gases. NOTE: CFC = chloro- fluorocarbon. SOURCE: National Research Council (1992). clude a variety of human activities that directly reduce habitat for endan- gered species. Many scientists believe that clearing of moist tropical for- ests is the leading cause of species extinction because of the extent of deforestation and because that biome is home to the greatest number of species found nowhere else. But the global rate and distribution of extinc- tions are still highly uncertain, as is the relative importance of different kinds of land use changes. Still, the analytical strategy of reasoning back from environmental effects to proximate causes remains viable. It will be necessary to revise the research agenda, of course, as knowledge im- proves about precisely which human activities are the major proximate causes of particular environmental effects. Estimating the magnitude of the main proximate causes may require 

128 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION difficult conceptual, analytical, or empirical work. To estimate strato- spheric ozone depletion, we need information on the ozone-depleting potential of a fairly large number of chemical compounds and their by- products over periods of time, as well as their rates of release. The most important proximate causes are identified by conducting detailed calcula- tions: they are those activities that directly result in releases with the greatest total ozone-depleting potential for a time horizon of concern. Analysis of the proximate causes of loss of biodiversity is much more difficult. It requires developing acceptable definitions and measures of biodiversity (e.g., should it be defined only as extinctions of species or also in other ways, such as loss of ecological niches or of genetic diversity within species; if the other concepts are important, how can they be mea- sured?). This analysis also requires studies of the effects of particular kinds of human activities on each index of biodiversity. Classifying hu- man activities for this purpose is not straightforward. For example, land use conversion does not have a unitary effect on biodiversity. Its impact depends on the ecosystem in which the land is converted, the new uses to which it is put, and also on such factors as the size, shape, and geographic relationships among unchanged parcels and those that are put to new uses. Because of such complexities, it may be misleading or inappropriate to measure threats to biodiversity in terms of a hypothetical average hect- are of land conversion. Determining which human activities have major proximate effects on environmental systems can point social and behavioral researchers to the broad types of activity that are likely to be most important in terms of a particular environmental change. For example, once the ozone-depleting potentials and the rates of emission of the various ozone-depleting com- pounds were calculated, it became possible to calculate the overall impact of each gas on ozone depletion, to consider the human activities that release each gas, and thus to identify the most ozone-depleting activities. This kind of analysis is important for choosing which kinds of consump- tion to study. Without this sort of analysis, behavioral researchers have often expended their efforts on studying human activities that are plausi- bly environmentally relevant and that they could easily address with concepts from their disciplines, but that did not contribute greatly to the environmental problem at hand (Stern and Oskamp, 1987). An alternative procedure for identifying important proximate causes of environmental change is to focus first on major materials and energy transformations and to estimate the environmental impacts of each. This is the approach used by Wernick and Allen in their contributions to Chap- ter 3. This approach is useful for studying the human causes of environ- mental conditions that have been linked to particular transformations of materials or energy. For example, Allen’s approach can, in principle, be 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 129 used to examine any environmental effect that has been attributed to a class of chemical compounds by analyzing flows of the particular com- pounds most strongly implicated in producing the effect. The difficulty in practice is to identify all the compounds with important direct or indi- rect effects. Analysis of the human activities that result in production of the responsible chemicals may then identify the most important targets for reducing the environmental effect. For ecological affects, such analy- sis may need to look not just at the actions that introduce a chemical into the biosphere but more specifically at the actions that introduce it in bioavailable forms. For example, most of the lead that human activity in the United States introduces into the environment is contained in motor vehicle batteries, and the great majority of this is repeatedly recycled. The far smaller amounts of lead in paint, gasoline, and shot have had more serious impacts because of their bioavailability. Disaggregating the Proximate Causes When important proximate causes have been identified, the next ana- lytic step is to disaggregate them to reflect the relative contribution of the many actors and activities that are responsible. There may be many sen- sible ways to disaggregate a particular proximate cause. Fossil fuel burn- ing can be subdivided according to parts of the world (countries or re- gions), economic sectors (transportation, industrial, commercial, etc.), immediate purposes (space heating, moving people or freight, generating electricity, etc.), types of actor (households, firms, government), or in other ways, depending on the analytic purpose. For important proximate causes, it may be useful to disaggregate in more than one way, creating categories that involve several types of division (e.g., transportation to work, commercial lighting). An impact criterion can help guide decisions about how far to disaggregate. That is, it makes sense to use finer divi- sions for types of actions or actors associated with major environmental impacts. It follows that the appropriate disaggregation will depend on the society being examined. Disaggregation can be represented in a tree diagram, as Figure 5-2 shows for greenhouse gas emissions, or in a table. Table 5-1 disaggre- gates carbon dioxide emissions in the United States by economic sector and purpose within each sector. It shows why, from the standpoint of climate change, much more analytic attention is warranted for under- standing the purchase and use of automobiles and light trucks (i.e., per- sonal transportation) than for understanding the purchase and use of water heating systems for buildings. It can be useful to disaggregate certain activities even further than shown in the table, as a way to direct researchers’ attention. For example, 

130 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION TABLE 5-1 Disaggregation of Carbon Dioxide Emissions by Economic Sector and End Use (percentages, United States, 1987) Economic Sector (%) End Use Industrial Buildings Transportation Total (%) Steam power, motors, appliances 19 7 26 Personal transportation (automobiles, light trucks) 20 20 Space heating 1a 16 17 Freight transport (heavy truck, rail, ship, other) 7 7 Heating for industrial processes 6 6 Lighting 1a 5 6 Cooling 0a 5 5 Air transporting 5 5 Water heating 3 3 Other 5 5 Total 32 36 32 100 NOTE: U.S. data are unrepresentative of world energy use in various ways. However, the United States is responsible for approximately 20 percent of global CO2 emissions. aTwo percent in the single category of heating, ventilating, air conditioning, and lighting was allocated 1 percent each to heating and lighting. SOURCE: National Research Council (1992). psychological research on energy use in the late 1970s was mainly ad- dressed to the study of daily behavior in residences, such as turning off lights in unoccupied rooms and altering thermostat settings to save on heating and cooling. An analysis that disaggregated residential energy use according to the energy-saving potential of different household ac- tivities showed, however, that infrequent purchase decisions, such as of insulation, furnaces, and major appliances, had significantly greater in- fluence on household energy use than the parallel daily behaviors involv- ing thermostats and appliance use (Stern and Gardner, 1981). The analy- sis directed behavioral scientists toward the study of a major class of environmentally significant household behavior. Research to disaggregate the proximate causes has proceeded farthest in the area of energy consumption (the contribution of Schipper in Chap- ter 3 illustrates some of what can be learned). The strategy of disaggrega- tion can also be applied to other environmentally significant consump- 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 131 tion. For instance, in the study of ozone depletion, the use of a particular ozone-depleting gas, such as the chlorofluorocarbon CFC-12, can be dis- aggregated to reflect its major uses in aerosols, refrigerants, and foams, as well as across countries. If releases of all the ozone-depleting gases were disaggregated in the same way, it would be possible to determine the proportion of total ozone-depleting potential attributable to particular activities such as refrigeration. Identifying and Analyzing the Driving Forces The next analytic step is to examine the major classes of environmen- tally significant action to identify their driving forces. This kind of work is part of the normal province of social and behavioral science, as its aim is to understand the causes of human activity (in this instance, activities that proximately change the environment). These driving forces may be of many types, the main classes of which are human population growth, levels of economic activity, technologies used in actions that affect the environment, political and economic institutions affecting action, and in- dividuals’ attitudes and beliefs (National Research Council, 1992). In Chapter 4, Dietz and Rosa examine the first two of these and suggest that the study of what these do not explain may reveal the roles of the other driving forces. Understanding the driving forces of an environmentally important activity usually presents serious analytical challenges. One reason, as noted in Chapter 1, is that the driving forces act together, often in nonad- ditive and nonlinear ways. They can affect the proximate causes of envi- ronmental change directly and also indirectly by acting on other driving forces. For example, when new chlorofluorocarbon-based technology made air conditioning practical for small commercial and residential uses after 1930, the technological advance set numerous social processes in motion, including large-scale population migrations to hotter climates in the United States, which led to increased use of air conditioning in homes far beyond what would have occurred without the migration. In this way, migration multiplied the effect of technology. In addition, the mi- gration increased fossil fuel consumption for transportation, as the new settlements were less dense geographically than the ones from which people migrated and therefore increased the demand for motorized travel while making mass transit less feasible economically. The migration also shifted the balance of representation in the U.S. Congress in favor of regions more dependent on air conditioning and automobiles, a change that may have made it more difficult to get political support for policies to increase the cost of energy consumption (see National Research Council, 1992:54-60). Thus, the environmental effect of technology was both direct 

132 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION and indirect through such variables as population change and policy. This example illustrates how the driving forces can influence each other: any single factor that seems to explain anthropogenic environmental change is likely to be affected by various other driving forces as well. This is one reason it is difficult to disentangle the causation of environmentally significant consumption. A second analytic difficulty is that the forces that drive environmen- tal change also respond dynamically to that change. For instance, intensi- fied agricultural production can increase soil erosion or pest infestations, which reduce agricultural productivity until effective changes are made in managing the environmental changes that the agricultural practices created. A third difficulty is that the relationships among the driving forces generally vary between places and over time. Deforestation has had a different collection of causes in different countries and at different times (Williams, 1990). The long-term environmental effects of a technological development can be very different from the short-term effects. The use of chlorofluorocarbons for refrigeration, for example, was an environmental and public health benefit at first because it replaced ammonia and other toxic or explosive refrigerants. The negative effect on stratospheric ozone became detectable only over a longer time scale and after the level of use had greatly increased. Although the task of untangling the causality of social phenomena such as those of consumption is difficult, it is a familiar one to social scientists. Experience suggests that the relative importance of the driving forces is likely to be quite situation-specific, even for relatively well-un- derstood activities like those involved in fossil energy use within a single sector of a single economy. For example, the factors determining invest- ment in energy-efficient technologies for the U.S. residential sector are different for occupant-owned and rental housing, for lower- and upper- income households, and probably also for investments in wall insulation and in energy-efficient appliances (e.g., Socolow, 1978; Ruderman et al., 1987; Stern et al., 1986). Research that begins by building relatively nar- row areas of knowledge about particular temporal and spatial scales is a first step toward developing a body of contingent generalizations that can make sense of broader bodies of data. ENVIRONMENT-FIRST STRATEGY AND OTHER APPROACHES The above discussion and several of the contributions to Chapters 3 and 4 illustrate the value of an environment-first strategy for directing research. Because this strategy begins with important environmental 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 133 changes and only later ventures into the social phenomena that may pro- vide deeper explanations, following it has a high probability of yielding results that meet a criterion of environmental importance. The environ- ment-first strategy is also valuable because it can help direct social scien- tists’ attention toward human activities and choices that, from an envi- ronmental standpoint, are particularly important to understand. As the contributions to Chapters 3 and 4 illustrate, there are two variants of this strategy. One works backward from transformations of materials and energy to their proximate causes in human activity and then to their driving forces. Wernick’s analysis (Chapter 3) of materials transforma- tions exemplifies this approach. The other begins with human activities, assesses their importance for particular environmental conditions, and if the activities are sufficiently important environmentally, proceeds to analysis of the proximate causes. The work of Schipper (Chapter 3) on travel and Lutzenhiser (Chapter 4) on home energy use are examples. Despite the strengths of the environment-first research strategy, other research strategies should also be pursued because they may be able to develop environmentally important knowledge that is not likely to emerge from environment-first research. One useful strategy is to begin with possible policy interventions. This strategy is exemplified by re- search on the effectiveness of price signals, information, and regulatory constraints for reducing consumer demand for environmentally damag- ing goods and services or for changing producers’ behavior so as to re- duce the environmental impact per unit of output. This strategy can have great practical value: for policy purposes, importance is not determined by which human activities do the most environmental damage but by the ones that can yield the greatest environmental improvement in response to practicable interventions. Research on intervention techniques helps determine what is practicable and cost-effective, identifies the possible secondary effects of interventions and the tradeoffs involved, and con- tributes to general knowledge about how environmentally significant human activities change. Because this strategy is most often employed to evaluate ways to change consumption, most of its applications are out- side the scope of this volume, and therefore no examples are included. Another alternative to looking at environment first, and one that is particularly relevant to the question of what drives environmentally im- portant consumption, is to focus first on social processes and then exam- ine their environmental effects. In Chapter 4, the reports by Wilk and by Kempton and Payne are examples. Wilk asks whether emulation of West- ern styles of consumption drives the styles adopted by people in develop- ing countries. Although the evidence so far is inconclusive, the question is environmentally important. Kempton and Payne consider the impacts of grand transformations in human history both on the environment and 

134 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION on the quality of human life. In doing this, they raise the question of what might be accomplished by future major social transformations. There are, of course, innumerable social processes that could be ex- amined for their environmental impacts. The social-process-first research strategy deserves special attention because it is the typical approach so- cial scientists take when they begin to work on environmental problems. Social researchers are likely first to try to understand anthropogenic envi- ronmental changes as resulting from the social processes they understand best. As the above examples suggest, this approach can identify some potentially important driving forces of environmental change that the environment-first strategy would be unlikely to uncover. However, the results from research following this strategy should be carefully judged against a criterion of environmental importance. Experience suggests that although the society-first research strategy can yield fresh insights, it is also likely to produce many findings of limited or uncertain environmental importance. One reason is that the environmental importance of a social phenomenon such as emulation, international trade, economic inequality, or slow transformations of basic social values cannot be estimated until careful analysis has been done. This analysis requires a long time series of multiple variables starting in the past and is likely to reveal that some of the driving forces have only limited environmental effects. Another reason is that it is difficult to evaluate the environmental significance of a single driving force indepen- dently of other driving forces. Because of the interactions of the driving forces and their interplay over time, it is inherently difficult to demon- strate the environmental importance of any indirect cause of environmen- tal change, particularly one acting over long time periods. Generally, it is easier to defend importance claims for driving forces that are linked to environmental effects by relatively short causal chains. For example, Schipper and his colleagues (1989) have argued that in- creases in female labor force participation in affluent countries increase fossil-energy consumption because they increase the proportion of women holding drivers’ licenses, the number of automobiles owned per capita, and the distance traveled. He links changes in household size to residen- tial energy consumption through number of households and per capita volume of heated and cooled residential space. These are strongly plau- sible claims, but even these require careful analysis to support them. For example, the increased energy demand caused by the need to heat or cool more space is offset (though only to a small extent) by savings resulting from the space being unoccupied during working hours. Other claims about driving forces involve longer and more debatable causal chains. For instance, it has been claimed that increasing adoption of so-called postmaterialist values (Inglehart, 1990) and proenvironmental 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 135 worldviews (e.g., Dunlap and Scarce, 1991; Milbrath, 1984) are affecting and will continue to affect human activity in proenvironmental ways. This hypothesis depends on psychological links from values or world- views to attitudes and from attitudes to behavior, as well as on the pre- sumption that the specific behaviors that may flow from changed values or worldviews will significantly alter the proximate causes of environ- mental change. Each such link needs to be established to support the hypothesis of environmental importance; the evidence so far suggests that these relationships do exist, but that each of the links is loose (Gardner and Stern, 1996). Still longer chains of causation are implicit in hypotheses about the environmental importance of various global social changes, such as the globalization of mass media (see Wilk’s report in Chapter 4), increasing worldwide urbanization, the emergence of global markets and global communications, democratization, and the resurgence of expressions of cultural identity (see National Research Council, 1992:156-160). Some of these broad social changes may indeed be critically important to long- term environmental trends, but it is difficult to evaluate claims about their importance because the lines of causation are so indirect and, in many cases, because of the long time lags hypothesized between causes and effects. The first step in evaluating the environmental relevance of research into such claims is to make explicit the causal links and the proximate causes that the social process is hypothesized to influence. That done, it is useful to assess the plausibility and evidential base for each link. For example, the emulation of U.S. consumption patterns in developing countries that Wilk discusses, even if actually occurring, is environmentally important only to the extent that particular kinds of con- sumption that are environmentally destructive (e.g., home air condition- ing) are copied. The environment-first research strategy may converge with the soci- ety-first strategy, and such convergence is usually a sign of research progress. For example, Lutzenhiser’s data on energy use identify accul- turation of immigrants as a factor that affects energy use in a consistent way, even controlling for other factors. This finding suggests a way of improving energy-demand modeling by taking the effects of international migration into account. Similarly, Schipper’s findings on female labor force participation and household size suggest some refinements in en- ergy models. Lutzenhiser also finds racial and ethnic differences in con- sumption, other factors held constant. This finding suggests that a closer look at racial and ethnic subcultures of consumption might increase un- derstanding of the ways attitudes, beliefs, and patterns of living affect the links between affluence and environmental outcomes. 

136 ENVIRONMENTALLY SIGNIFICANT CONSUMPTION CONCLUSIONS Research on the environmental impacts of consumption should in- clude a mixture of investigations using environment-first, policy-oriented, and society-first approaches. Because it is anchored to environmental outcome variables of obvious importance, the environment-first strategy is likely to lead to cumulative research that will, over time, continually refine knowledge about the causes of significant anthropogenic environ- mental changes. This strategy will also increasingly clarify the human choices, actions, and actors that provide important targets for policy in- tervention, and it will improve the ability of analysts to make projections of anthropogenic inputs into biophysical systems. The policy-oriented strategy is likely to result in a different sort of cumulative knowledge, leading to increased understanding of the poten- tial for shaping the forces that drive environmental change. It will im- prove understanding of the conditions under which various kinds and combinations of policy interventions are effective and the extent to which an intervention that works with one kind of behavior or in one social context may be transferable elsewhere. In addition, it is likely to increase basic knowledge about change in individuals and social systems and thus contribute to social science generally as well as to policy analysis. The society-first strategy is much less likely to generate cumulative research on environmental conditions, but it has compensating advan- tages. It will occasionally reveal unsuspected but important anthropo- genic influences on the environment, especially indirect and long-term ones. Emulation may turn out to be such an influence. Another may be an apparent trend in some wealthy countries toward occupying leisure time by shopping. Yet another, with the potential to alter the environ- mental impact of human activity, is a trend in the United States for the management of local resources such a parklands and riverbanks to de- volve from centralized institutions to local ones—a change that may shift management priorities. Identifying such indirect or unsuspected influences on environmen- tal quality may help researchers who are following the environment-first strategy to reconsider and improve their conceptual models. Society-first research may also identify possibilities for interventions that long-range planners can use and that may be more socially acceptable than short- term interventions that appear to some people as drastic action to avert a projected emergency that they doubt will ever arise. Changes in the rules governing resource-management institutions could be such an example. The society-first approach is also likely to lead to insights of great value for constructing scenarios and projections of future human-envi- ronment interactions. In addition, it may help interest more social scien- 

STRATEGIES FOR SETTING RESEARCH PRIORITIES 137 tists in environmental issues by showing them clear connections between their normal research concerns and an important public policy issue. In- volvement by a greater variety of social scientists and collaboration with natural scientists can not only improve research on environment and con- sumption but also enrich the social science disciplines by bringing them into contact with new sets of phenomena. REFERENCES Dunlap, R.E., and R. Scarce 1991 The polls/poll trends: Environmental problems and protection. Public Opinion Quarterly 55:713-734. Gardner, G.T., and P.C. Stern 1996 Environmental Problems and Human Behavior. Needham Heights, Mass.: Allyn and Bacon. Inglehart, R. 1990 Culture Shift in Advanced Industrial Society. Princeton, N.J.: Princeton University Press. Milbrath, L. 1984 Environmentalists: Vanguard for a New Society. Albany: State University of New York Press. National Research Council 1992 Global Environmental Change: Understanding the Human Dimensions. P.C. Stern, O.R. Young, and D. Druckman, eds. Committee on the Human Dimensions of Global Change. Washington: National Academy Press. Ruderman, H., M.D. Levine, and J.E. McMahon 1987 The behavior of the market for energy efficiency in residential appliances includ- ing heating and cooling equipment. The Energy Journal 8:101-124. Schipper, L., S. Bartlett, D. Hawk, and E. Vine 1989 Linking life-styles and energy use: A matter of time? Annual Review of Energy 14:273-320. Socolow, R.H. 1978 The Twin Rivers program on energy conservation in housing: Highlights and conclusions. Energy and Buildings 1:313-324. Stern, P.C., E. Aronson, J.M. Darley, D.H. Hill, E. Hirst, W. Kempton, and T.J. Wilbanks 1986 The effectiveness of incentives for residential energy conservation. Evaluation Review 10:147-176. Stern, P.C., and G.T. Gardner 1981 Psychological research and energy policy. American Psychologist 36:329-342. Stern, P.C., and S. Oskamp 1987 Managing scarce environmental resources. Pp. 1043-1088 in I. Altman and D. Stokols, eds., Handbook of Environmental Psychology. New York: Wiley- Interscience. Williams, M. 1990 Forests. Pp. 179-201 in B.L. Turner II, W.C. Clark, R.W. Kates, J.F. Richards, J.T. Mathers, and W.B. Meyer, eds., The Earth as Transformed by Human Action. Cam- bridge, England: Cambridge University Press.