Analyzing Historical Contingency with Formal Methods

Abstract

Historical sociologists have been properly critical of the tendency within the social sciences to ignore temporality and, consequently, historical contingency in social processes. These critical assessments have generated alternative formal research methods that either (a) continue to use variable-centered procedures but allow relations between variables to change through time, or (b) abandon variable-centered procedures altogether and move exclusively to event-centered analyses. After considering alternative conceptions of historically contingent determination that may be produced by two different formal methods (time-varying parameter regression and event structure analysis), the authors design a combined strategy that temporally nests interdependent forms of contingency and illustrates the approach in an empirical analysis of the “relief explosion” in the late 1960s. Contrary to exhortations for exclusivist strategies, the results suggest that variable-centered and event-centered methods can be fruitfully combined. In particular, event structure analysis is shown to be a useful qualitative procedure for “unpacking” and interpreting quantitative turning points in time-varying parameter time series models. This strategy allows macroquantitative and historical/cultural processes to be integrated with formal methods.