Multiple Levels of Analysis and Time Scales

Abstract

Explanations for human behavior often involve factors operating at multiple levels of analysis (from individuals to organizations to nations) and across different time scales (immediate, developmental, and evolutionary). Continuing debates over methodological individualism within the social sciences ignore the growing consensus within philosophy of social science that influences on social life are present across individual and collective units, each with influential histories. Even if models of individual interaction offer useful templates, they do not imply that any social process started anew at any particular point in time or that individuals created social structures without being influenced by prior institutions. Using a comparison of traffic and weather, I find that these difficulties are no worse for social than natural science, once we welcome both simple patterns and complex multilevel processes. In both cases, much progress is made simply through observational generalization and many complaints involve our ability to change the world, not understand it.