This project was supported
by the Political Science program
of the National Science Foundation
Two datasets are available – a simple summary of the estimates and a larger dataset that includes 2,000 observations for each state and year. The larger dataset is required to replicate the confidence intervals in the figures. Both datasets were created in STATA 13.1
Each observation is a single state and election year. The dataset totals to 50 observations – 10 sample states observed in five presidential election years.
Estimates and measures of uncertainty
Each observation is one sample from the Markov Chain Monte Carlo simulation used to generate the underlying estimates that were used to calculate the state-level quantities. Each state-election year has 2,000 observations - 2,000 samples from the simulation after the chain had converged. The total dataset is 2,000 x 10 x 5 or 100,000 observations. We used this dataset to produce figures with the average for the state and a 90% confidence interval. This dataset can be used to replicate all of the figures in Chapters 2-8 and several figures in Chapter 9.
codebook.pdf : Variable names and descriptions (2pp.)