Noah Buckley
Papers Under Review and Conference Papers
- “Using Russian Internet Search Data for Measurement and Prediction”
- “Bayesian Hierarchical Modeling of Time-Series Cross-Sectional Political Data with Serial Correlation”
- In this paper I provide a systematic exploration of multilevel/hierarchical techniques for modeling time series cross-sectional (TSCS) data with serial correlation as most often encountered in quantitative political science. I first develop and present the hierarchical modeling with autoregressive distributed lag (HM-ADL) model for estimating TSCS data. I then employ Bayesian MCMC multilevel modeling to test the performance of this technique against a wide variety of alternatives. This includes evaluation of the random effects portion of the model when one would expect it to perform poorly, i.e., when the unobserved unit effects are correlated with a predictor. Using a flexible data generating process, I conduct several Monte Carlo experiments with a substantial set of experimental conditions to examine the performance of these techniques for TSCS analysis.