The Historical Experiment as a Research Strategy in the Study of World Politics

Abstract

As one looks back on the important developments in political science over the past two decades, there is much to be applauded. While it may be premature to say that we have come “of age” as a scientific discipline, the field is clearly in better shape today than it was in the early 1950s. One indicator is the ratio between mere speculation and observed empirical regularities reported in our journal articles. Another is the decline in the percentage (if not in absolute numbers) of our colleagues who insist that political phenomena are just not amenable to scientific examination. A third might be the dramatic increase in the number of political scientists who have been exposed to training in the techniques and rationale of data making and data analysis. The list could be extended, but we need not do so here.On the other hand, a stance of comfortable complacency would be very premature. Not only have we fared badly in coming to grips with the knowledge-action relationship in the abstract, but we have by and large done a poor job of shaping the policies of our respective national, provincial, and local governments. Since others as well as myself have dealt—if not definitively—with these issues before, let me eschew further discussion of the knowledgeapplicationquestion for the moment, and go on to matters of basic research. Of the more serious flaws to date, two stand out particularly. One is the lack of balance between a concern for cumulativeness on the one hand and the need for innovation on the other. My impression is that students ofnationalpolitics (at least those who work in the vineyard of empirical regularities) have been more than conscientious in staying with one set of problems, such as the relationship between political attitudes and voting behavior. But students ofinter-national politics have, conversely, tended to move all too quickly from one problem to another, long before cumulative evidence has been generated and before our findings are integrated into coherent wholes.