History, Historiography, and Political Science: Multiple Historical Records and the Problem of Selection Bias

Abstract

Social scientists who use history as a laboratory for theory development use the work of historians to construct background narratives which can then be coded according to theoretically relevant categories. Yet, virtually no attention has been paid to how these historical monographs are to be chosen. On most periods and themes of interest available accounts differ, not only substantively but also with respect to the implicit theories and conceptual frameworks used to establish salience or produce commonsensical explanations. Unself-conscious use of historical monographs thus easily results in selection bias. Social scientists are bound to be more attracted to and convinced by accounts that accord with the expectations about events contained in the concepts they deploy and the theories they seek to test. Consideration of recent developments in historiographical theory supports the argument that responsible techniques for using historical sources are available, but they require understanding the extent to which patterns within historiography, rather than “History,” must be the direct focus of investigation and explanation. Such an approach has the added advantage of helping to generate historically based studies where observations or cases outnumber variables.