The large majority of inferences drawn in empirical political research follow from model-based associations (e.g., regression). Here, we articulate the benefits of predictive modeling as a complement to this approach. Predictive models aim to specify a probabilistic model that provides a good fit to testing data that were not used to estimate the model’s parameters. Our goals are threefold. First, we review the central benefits of this under-utilized approach from a perspective uncommon in the existing literature: we focus on how predictive modeling can be used to complement and augment standard associational analyses. Second, we advance the state of the literature by laying out a simple set of benchmark predictive criteria. Third, we illustrate our approach through a detailed application to the prediction of interstate conflict.

Over the past decade, a number of new measures have been developed that attempt to capture the political orientation of both incumbent and nonincumbent candidates for Congress, as well as other offices, on the same scale. These measures pose the possibility of being able to answer a host of fundamental questions about political accountability and representation. In this paper, we examine the properties of six recent measures of candidates’ political orientations in different domains. While these measures are commonly viewed as proxies for ideology, each involves very different choices, incentives, and contexts. Indeed, we show that there is only a weak relationship between these measures within party. This suggests that these measures are capturing domain-specific factors rather than just candidates’ ideology. Moreover, these measures do poorly at distinguishing between moderate and extreme roll call voting records within each party. As a result, they fall short when it comes to facilitating empirical analysis of theories of accountability and representation in Congress. Overall, our findings suggest that future research should leverage the conceptual and empirical variation across these measures and avoid assuming they are synonymous with candidates’ ideology.

Missing outcome data plague many randomized experiments. Common solutions rely on ignorability assumptions that may not be credible in all applications. We propose a method for confronting missing outcome data that makes fairly weak assumptions but can still yield informative bounds on the average treatment effect. Our approach is based on a combination of the double sampling design and nonparametric worst-case bounds. We derive a worst-case bounds estimator under double sampling and provide analytic expressions for variance estimators and confidence intervals. We also propose a method for covariate adjustment using poststratification and a sensitivity analysis for nonignorable missingness. Finally, we illustrate the utility of our approach using Monte Carlo simulations and a placebo-controlled randomized field experiment on the effects of persuasion on social attitudes with survey-based outcome measures.

This article presents a new method and open source R package that uses syntactic information to automatically extract source–subject–predicate clauses. This improves on frequency-based text analysis methods by dividing text into predicates with an identified subject and optional source, extracting the statements and actions of (political) actors as mentioned in the text. The content of these predicates can be analyzed using existing frequency-based methods, allowing for the analysis of actions, issue positions and framing by different actors within a single text. We show that a small set of syntactic patterns can extract clauses and identify quotes with good accuracy, significantly outperforming a baseline system based on word order. Taking the 2008–2009 Gaza war as an example, we further show how corpus comparison and semantic network analysis applied to the results of the clause analysis can show differences in citation and framing patterns between U.S. and English-language Chinese coverage of this war.

Media-based event data—i.e., data comprised from reporting by media outlets—are widely used in political science research. However, events of interest (e.g., strikes, protests, conflict) are often underreported by these primary and secondary sources, producing incomplete data that risks inconsistency and bias in subsequent analysis. While general strategies exist to help ameliorate this bias, these methods do not make full use of the information often available to researchers. Specifically, much of the event data used in the social sciences is drawn from multiple, overlapping news sources (e.g., Agence France-Presse, Reuters). Therefore, we propose a novel maximum likelihood estimator that corrects for misclassification in data arising from multiple sources. In the most general formulation of our estimator, researchers can specify separate sets of predictors for the true-event model and each of the misclassification models characterizing whether a source fails to report on an event. As such, researchers are able to accurately test theories on both the causes of and reporting on an event of interest. Simulations evidence that our technique regularly outperforms current strategies that either neglect misclassification, the unique features of the data-generating process, or both. We also illustrate the utility of this method with a model of repression using the Social Conflict in Africa Database.

What explains why some survey respondents answer truthfully to a sensitive survey question, while others do not? This question is central to our understanding of a wide variety of attitudes, beliefs, and behaviors, but has remained difficult to investigate empirically due to the inherent problem of distinguishing those who are telling the truth from those who are misreporting. This article proposes a solution to this problem. It develops a method to model, within a multivariate regression context, whether survey respondents provide one response to a sensitive item in a list experiment, but answer otherwise when asked to reveal that belief openly in response to a direct question. As an empirical application, the method is applied to an original large-scale list experiment to investigate whether those on the ideological left are more likely to misreport their responses to questions about prejudice than those on the right. The method is implemented for researchers as open-source software.

Do natural disasters help or hurt politicians’ electoral fortunes? Research on this question has produced conflicting results. Achen and Bartels (2002, 2016) find that voters punish incumbent politicians indiscriminately after such disasters. Other studies find that voters incorporate the quality of relief efforts by elected officials. We argue that results in this literature may be driven, in part, by a focus on contemporary cases of disaster and relief. In contrast, we study a case of catastrophic flooding in the American South in 1927, in which disaster aid was broadly and fairly distributed and Herbert Hoover (the 1928 Republican presidential candidate) was personally responsible for overseeing the relief efforts. Despite the distribution of unprecedented levels of disaster aid, we find that voters punished Hoover at the polls: in affected counties, Hoover’s vote share decreased by more than 10 percentage points. Our results are robust to the use of synthetic control methods and suggest that—even if voters distinguish between low- and high-quality responses—the aggregate effect of this disaster remains broadly negative. Our findings provide some support for Achen and Bartels’ idea of blind retrospection, but also generate questions about the precise mechanisms by which damage and relief affect vote choice.

An influential paper by Caughey and Sekhon (2011a) suggests that the outcomes of very close US House elections in the postwar era may not be as-if random, thus calling into question this application of regression discontinuity for causal inference. We show that while incumbent party candidates are more likely to win close House elections, those who win are no different on observable characteristics from those who lose. Further, all differences in observable characteristics between barely winning Democrats and barely winning Republicans vanish conditional on which party is the incumbent. Any source of a special incumbent party advantage in close elections must be due to variables that cannot be observed. This finding supports the conclusion of Eggers et al. (2015) that Caughey and Sekhon’s discovery of lopsided wins by incumbents in close races is a mere statistical fluke.