to learn from the substantial experience of designing and implementing environmental policy in the United States, and to anticipate environmental trends and future policy needs. By focusing scientific efforts increasingly on decision relevance, such a program of measurement, evaluation, and analysis would increase the influence of empirical evidence and empirically supported theory in environmental decisions relative to the influences of politics and ideology. It would integrate the social sciences and the natural sciences of the environment and build a knowledge base that would better inform practical decision making while also informing scientific research.

Processes for determining which research is most decision-relevant should be participatory. Choices about how to construct indicators, evaluate policies, and so forth should be made with the participation of the full range of likely users of measures, evaluations, and analyses. This approach has previously been recommended for shaping scientific research agendas in a number of disparate areas of environmental science (e.g., National Research Council, 1996, 1999d, 2003, 2004c), and we state it here as a general principle. Broad involvement is essential to enhance the decision relevance and credibility of measures, evaluations, and analyses. Choices about measurement are not purely technical, so they are not appropriately left to analysts alone. Measurement focuses attention on what has been measured and away from what is unmeasured, thus embodying values about what is most worthy of attention. The affected parties in environmental decisions often disagree about what is most worth measuring, which outcomes of policies are most important, and the like, and the range of measures needed to make an analysis credible may not be obvious to scientists or government officials unless the various potential users of indicators are involved. Choices about what evidence to collect for policy are probably most appropriately made through broad-based analytic-deliberative processes, such as described in Chapter 2.

Government should implement the strategy of seeking decision relevance in each area of environmental policy. This will require four kinds of research activities: (1) developing decision-relevant indicators, (2) evaluating past policies and programs, (3) improving the scientific capability to anticipate future environmental conditions and problems, and (4) measuring and monitoring the distribution of environmental impacts in the population in relation to issues of environmental inequities and their abatement. Within each of these activities, decisions about research priorities should be informed by dialogue between the potential producers and the potential users of the research.

Evidence-based environmental policy depends on having measures and