Observational political science overwhelmingly involves phenomena that occur and change over time. Temporal dependencies abound both between and within many social processes. Of course, failing to account for temporal dependencies in dynamic data violates the classical regression assumptions. Yet unfortunately some analysts seem to view the dynamic processes in their data as only a statistical nuisance to be fixed. Modeling, rather than merely correcting, these dynamic processes, offers potentially rich new inferences regarding fundamental problems in the discipline. Yet properly modeling dynamic processes requires a firm understanding of both the foundations of, and the frontiers of research in, time series analysis. To that end, we are pleased to announce the publication of a new book, Time Series Analysis for the Social Sciences.
Utilizing practical examples from the fields of international conflict, political economy, political behavior, democratic accountability, criminology, and sociology, we present a comprehensive introduction to time series analysis. This introduction is appropriate for researchers and graduate students with at least one course in multivariate regression.
We begin with foundational material, covering univariate time series analysis utilizing the Box-Jenkins approach, along with time series analysis within a classical regression framework. We continue with an exposition of multivariate time series analysis, focusing first on vector autoregression. We then move to diagnosing and modeling with nonstationary (unit root) data. This discussion naturally leads into a presentation of cointegration analysis and the error correction model. Finally, we present an overview of several advanced techniques: forecasting, modeling volatility, fractional integration, and estimating and modeling with multiple unknown structural breaks. Each chapter concludes with an overview of recent relevant literature on each specific topic for interested readers. A supplementary appendix covers the calculus of difference equations for readers interested in the mathematical logic underpinning the techniques covered. Finally, a companion website offers both data and STATA replication code for all the examples in the book. We hope this book proves useful to researchers across the social sciences.