A Primer on Statistical Analysis of Dynamical Systems in Historical Social Sciences (with a Particular Emphasis on Secular Cycles)
This primer explicates the conceptual foundations of the statistical approach to detecting dynamical feedbacks. It is assumed that we have time-series data on several aspects of the studied system. The basic idea of the approach is to regress discrete rates of change of measured variables on variables themselves. I discuss several issues involved in the analysis, such as how to select the appropriate time step, or the delay parameter. The goal of the analysis is to determine whether a particular predictor variable, or set of variables, has a statistically detectable effect on the response. This is accomplished by cross-validation.