Conceptualizing and measuring environmentalism
It is well established that the scale of human economic activities, typically thought of as the product of population and affluence (P*A), is a major driver of human stress on the environment (Rosa and Dietz 2012). Much work remains to be done to unpack how demographic and economic scale influence environmental stress. But we are interested in understanding better the factors that might compensate for the effects of scale, particularly since for at least much of the remainder of the 21st century we can anticipate scale will increase globally and in most nations and subnational units. Shwom (Shwom 2011), in an influential argument, points our attention to politics as well as to the importance of business decision making in shaping environmental stress. Further, research on environmentally significant consumption shows that household decisions about consumption can be environmentally consequential (Dietz, Gardner, Gilligan, Stern, and Vandenbergh 2009).
This suggests to us that we need a fairly broad conceptualization of environmentalism. Certainly the “new wave” of environmentalism that emerged circa the first Earth Day focused on individuals as both consumers and as citizens; on the actions for better and for worse of businesses; and on policy, regulation and other government action. Thus in attempting to assess the effects of environmentalism as a potential moderator to the stresses that come from scale, we want a broad approach, one that includes in environmentalism not only the strength and effectiveness of formal environmental movement organizations, but also the values, beliefs, norms and decisions of individuals as consumers and as citizens, the actions of firms, and not only the existence of government policies and programs but also their effectiveness. All of these things could, in principle, moderate the stress placed on the environment.
An ideal data set would include measures of each of these factors and allow us to estimate the relative importance of each in reducing stress on the environment. But as too often has been the case in the environmental social sciences, our ability to conduct the analyses implied by our conceptualizations is constrained by limits in what data are available. While there are many compelling reasons to use U.S. states as a unit of analysis, there is a paucity of indicators of environmentalism that are available for all states over time.
Congressional voting as a measure of state politics.
For this analysis we have chosen to use the score assigned by the League of Conservation Voters to state Congressional delegations as our measure of state environmentalism. Since many readers are not familiar with the logic of indicators, we want to explain our reasoning in a bit more detail than was possible within the space constraints of the paper.
Why Congressional voting? There are two justifications for using Congressional voting as an indicator of state politics. The first is common sense. Members get to Congress because of the support of the electorate, but also of political and economic elites who provide funds and otherwise engage in the political process. Thus who gets to Congress and how they vote once there are shaped by the views of both the public and political elites. So we would expect that the votes of a Congressional delegation to a substantial degree reflect the politics of the state. (Granted the can be considerable diversity in state politics, but given there are at least three members of Congress from every state, the average score on voting for the state can capture both uniformity (for example especially high or low scores) or diversity in state politics (high and low scores of individual members average out into moderate scores).)
The second rationale is grounded in the literature on state politics. A substantial and respected body of research has examined the link between Congressional voting and state politics (Berry, Fording, Rinquist, Hanson, and Klarner 2010; Berry, Ringquist, Fording, and Hanson 1998). The voting patterns of state Congressional delegations as scored by interest organizations with clear ideological positions have repeatedly proven useful in analyses of state politics. Thus we build on this literature that demonstrates the utility of Congressional voting as a measure of state politics, and simply extend it to environmentalism.
Why the League of Conservation Voters score? The League of Conservation Voters (LCV)was founded in 1969. Since 1970 its “Scorecard” rates each member of Congress each year on a scale from 0-100 based on their voting record. LCV staff in consultation with other experts select key votes in the House and the Senate and develop the score for each member based on those votes. This practice is essentially identical to the approach used in developing scores by Americans for Democratic Action and the Committee on Political Education of the AFL-CIO, the scores used in the state politics literature in using Congressional voting to assess state politics. The LCV is the only consistent source of scores on environmental votes in Congress of which we are aware.
What does it mean?
We are using the LCV scores for the state Congressional delegation as an indicator of state environmentalism, paralleling the use of such scores in analyses of state politics. We are not arguing that the votes in Congress somehow directly influence state greenhouse gas emissions. Rather, our logic is essentially a latent variable model with only one indicator of the latent variable. This might diagrammed as below. The circle represents the latent variable we do not observe directly, environmentalism. The two squares represent the two variables on which we have data. Since we don’t have data on environmentalism directly, we treat Congressional voting as an indicator of environmentalism and examine its link to CO2 emissions to assess the relationship between the latent variable, environmentalism, and emissions. If we had more than one indicator of environmentalism for states over time, we could use structural equation modeling techniques for multiple indicators to account for the reliability with which we are measuring environmentalism.[1] However, since measurement error tends to reduce the ability to detect significant effects in most circumstances, and since we find rather strong effects of environmentalism on emissions using our single indicator, it is likely that if we had multiple indicators we would find even stronger effects.
Having multiple indicators of environmentalism, in addition to reducing measurement error, would allow us to estimate the effects of each aspect of environmentalism on emissions. Thus we would be able to disentangle the effects of individual behavior as consumers and as citizens, social movement organizations, business support for environmental protection and both the existence of state and local policies and programs and the enthusiasm and effectiveness with which they are pursued.
Berry, William D, Richard C Fording, Evan J Rinquist, Russell L Hanson, and Carl Klarner. 2010. "Measuring Citizen and Government Ideology in the American States: A Re-appraisal " State Politics and Polity Quarterly 10:117-135.
Berry, William D, Evan J Ringquist, Richard C Fording, and Russell L Hanson. 1998. "Measuring Citizen and Government Ideology in the American States, 1960-1993." American Journal of Political Science 42:327-348.
Dietz, Thomas, Gerald T. Gardner, Jonathan Gilligan, Paul C. Stern, and Michael P. Vandenbergh. 2009. "Household Actions Can Provide a Behavioral Wedge to Rapidly Reduce U.S. Carbon Emissions." Proceedings of the National Academy of Sciences 106:18452-18456.
Rosa, Eugene A. and Thomas Dietz. 2012. "Human drivers of national greenhouse-gas emissions." Nature Clim. Change 2:581-586.
Shwom, Rachael L. 2011. "A middle range theorization of energy politics: The U.S. struggle for energy efficient appliances." Enviromental Politics 20:705-726.
[1]In fact our score is the average of the House average and the Senate average for a given state in a given year. If we take these as two indicators of environmentalism, they are correlated r=0.691 which is equivalent to an alpha reliability of 0.809.