Jurisprudential regime theory is a legal explanation of decision-making on the U.S. Supreme Court that asserts that a key precedent in an area of law fundamentally restructures the relationship between case characteristics and the outcomes of future cases. In this article, we offer a multivariate multiple change-point probit model that can be used to endogenously test for the existence of jurisprudential regimes. Unlike the previously employed methods, our model does so by estimating the locations of many possible change-points along with structural parameters. We estimate the model using Markov chain Monte Carlo methods, and use Bayesian model comparison to determine the number of change-points. Our findings are consistent with jurisprudential regimes in the Establishment Clause and administrative law contexts. We find little support for hypothesized regimes in the areas of free speech and search-and-seizure. The Bayesian multivariate change-point model we propose has broad potential applications to studying structural breaks in either regular or irregular time-series data about political institutions or processes.

Social scientists rely on surveys to explain political behavior. From consistent overreporting of voter turnout, it is evident that responses on survey items may be unreliable and lead scholars to incorrectly estimate the correlates of participation. Leveraging developments in technology and improvements in public records, we conduct the first-ever fifty-state vote validation. We parse overreporting due to response bias from overreporting due to inaccurate respondents. We find that nonvoters who are politically engaged and equipped with politically relevant resources consistently misreport that they voted. This finding cannot be explained by faulty registration records, which we measure with new indicators of election administration quality. Respondents are found to misreport only on survey items associated with socially desirable outcomes, which we find by validating items beyond voting, like race and party. We show that studies of representation and participation based on survey reports dramatically misestimate the differences between voters and nonvoters.

Political scientists use randomized treatment assignments to aid causal inference in field experiments, psychological laboratories, and survey research. Political research can do considerably better than completely randomized designs, but few political science experiments combine random treatment assignment with blocking on a rich set of background covariates. We describe high-dimensional multivariate blocking, including on continuous covariates, detail its statistical and political advantages over complete randomization, introduce a particular algorithm, and propose a procedure to mitigate unit interference in experiments. We demonstrate the performance of our algorithm in simulations and three field experiments from campaign politics and education.

In this article, we present a technique and critical test statistic for assessing the fit of a binary-dependent variable model (e.g., a logit or probit). We examine how closely a model's predicted probabilities match the observed frequency of events in the data set, and whether these deviations are systematic or merely noise. Our technique allows researchers to detect problems with a model's specification that obscure substantive understanding of the underlying data-generating process, such as missing interaction terms or unmodeled nonlinearities. We also show that these problems go undetected by the fit statistics most commonly used in political science.

The ability to veto legislation is the most important formal power of the presidency in the legislative process, and presidents' veto behavior has thus attracted a great deal of theoretical interest. Game-theoretic models of congressional—presidential interactions explain vetoes by incorporating incomplete information over the distribution of presidential or congressional types. In this article, I examine two prominent such theories, as well as a simple “noisy” extension of a complete information theory. I show that each makes strong predictions not only about vetoes themselves but also about the resulting override votes; given that overrides are so closely connected with vetoes, any valid theory of the latter must be able to successfully explain the former. I test these predictions empirically and show that support for each of these theories in presidential veto and override data from 1973 to 2008 is quite weak. This negative result suggests that current models of the veto are incomplete; I sketch some possibilities for extension in the conclusion.

“Testing Theories of Congressional-Presidential Interaction with Veto Override Rates” (henceforth “Veto Override Rates”) offers several tests of two models of vetoes and finds the models wanting. The paper concludes that something is seriously amiss with the models. In my view, the problem lies not in the models but in the tests. Understanding why the tests miss the mark is helpful in understanding models of veto politics, and more generally in thinking about testing strategies when multiple models analyze different causal mechanisms that hold under different circumstances. I should note immediately that the effort in the paper to think hard about override rates is admirable; it simply does not go far enough.