Strategies for Collating Diverse Scientific Evidence in the Analysis of Population Health Characteristics

Abstract

The analysis of many social and health policy issues requires the use of multiple data sources from a diverse body of scientific and technical studies. Although individual data sets are rigorously analyzed, integration of the results of these analyses to resolve policy questions is often accomplished by informal or subjective strategies based on procedures designed to generate consensus among scientific experts. In this article we discuss a model for conducting a more formal integration of multiple data sources (including subjective or theoretical judgments). The advantages of such models over consensus generation by informal means are that (1) they can produce very specific quantitative measures of the implications of alternative policies; (2) their assumptions are more readily reviewable; (3) they can be validated against data; and (4) they formally link experimental and survey data, organizing our knowledge base so that priorities for improving the knowledge base can be determined systematically. These advantages suggest that the use of formal models can be a valuable adjunct to informal consensus-generating procedures. An example of how such a modeling strategy is applied to the monitoring of population health is presented and discussed.