It is very difficult to know how international social linkages affect domestic ideological polarization because we can never observe polarization occurring both with and without international connections. To estimate this missing counterfactual, we employ a new statistical method based on Bayesian item-response theory that permits us to disaggregate polarization after the Arab Uprisings into domestic and transnational components. We collected a dataset of Twitter accounts in Egypt and Tunisia during the critical year of 2013, when the Egyptian military overthrew the Islamist President Mohamed Morsi. We find that the coup increased retweets among Egyptian ideological allies by 50% each day following the coup and decreased cross-ideological retweets by 25%. Tunisian Twitter communities also showed stronger intragroup retweeting although at lower levels than in Egypt. Counter-intuitively, our model shows that the additional polarization in Tunisia after the coup appears to have dampened further polarization among Islamists in Egypt.

We model the conditional distribution of seats given vote shares induced by national electoral systems using a stochastic threshold of representation and a disproportionality parameter that regulates allocation for parties above the threshold. We establish conditions for the parameters of this model to be identified from observed seats/votes data, and develop a Maximum a Posteriori Expectation-Maximization (MAP-EM) algorithm to estimate them. We apply the procedure to 116 electoral systems used in 417 elections to the lower house across 36 European countries since WWII. We reject a test of model fit in only 5 of those systems, while a simpler model without thresholds is rejected in favor of our estimated model in 49 electoral systems. We find that the two modal electoral system configurations involve higher thresholds with seat allocation for parties exceeding thresholds that does not statistically differ from perfectly proportional allocation (32.76% of all systems); and systems for which we cannot reject the absence of a national threshold but exhibit disproportional seat allocation for parties eligible for seats (38.79% of all systems). We also develop procedures to test for significant changes in electoral institutions and/or the distribution of seats.

In this article, we present a machine learning-based solution for matching the performance of the gold standard of double-blind human coding when it comes to content analysis in comparative politics. We combine a quantitative text analysis approach with supervised learning and limited human resources in order to classify the front-page articles of a leading Hungarian daily newspaper based on their full text. Our goal was to assign items in our dataset to one of 21 policy topics based on the codebook of the Comparative Agendas Project. The classification of the imbalanced classes of topics was handled by a hybrid binary snowball workflow. This relies on limited human resources as well as supervised learning; it simplifies the multiclass problem to one of binary choice; and it is based on a snowball approach as we augment the training set with machine-classified observations after each successful round and also between corpora. Our results show that our approach provided better precision results (of over 80% for most topic codes) than what is customary for human coders and most computer-assisted coding projects. Nevertheless, this high precision came at the expense of a relatively low, below 60%, share of labeled articles.

One of the strongest findings across the sciences is that publication bias occurs. Of particular note is a “file drawer bias” where statistically significant results are privileged over nonsignificant results. Recognition of this bias, along with increased calls for “open science,” has led to an emphasis on replication studies. Yet, few have explored publication bias and its consequences in replication studies. We offer a model of the publication process involving an initial study and a replication. We use the model to describe three types of publication biases: (1) file drawer bias, (2) a “repeat study” bias against the publication of replication studies, and (3) a “gotcha bias” where replication results that run contrary to a prior study are more likely to be published. We estimate the model’s parameters with a vignette experiment conducted with political science professors teaching at Ph.D. granting institutions in the United States. We find evidence of all three types of bias, although those explicitly involving replication studies are notably smaller. This bodes well for the replication movement. That said, the aggregation of all of the biases increases the number of false positives in a literature. We conclude by discussing a path for future work on publication biases.

Economics games such as the Dictator and Public Goods Games have been widely used to measure ethnic bias in political science and economics. Yet these tools may fail to measure bias as intended because they are vulnerable to self-presentational concerns and/or fail to capture bias rooted in more automatic associative and affective reactions. We examine a set of misattribution-based approaches, adapted from social psychology, that may sidestep these concerns. Participants in Nairobi, Kenya completed a series of common economics games alongside versions of these misattribution tasks adapted for this setting, each designed to detect bias toward noncoethnics relative to coethnics. Several of the misattribution tasks show clear evidence of (expected) bias, arguably reflecting differences in positive/negative affect and heightened threat perception toward noncoethnics. The Dictator and Public Goods Games, by contrast, are unable to detect any bias in behavior toward noncoethnics versus coethnics. We conclude that researchers of ethnic and other biases may benefit from including misattribution-based procedures in their tool kits to widen the set of biases to which their investigations are sensitive.

Citizens’ beliefs about uncertain events are fundamental variables in many areas of political science. While beliefs are often conceptualized in the form of distributions, obtaining reliable measures in terms of full probability densities is a difficult task. In this letter, we ask if there is an effective way of eliciting beliefs as distributions in the context of online surveys. Relying on experimental evidence, we evaluate the performance of five different elicitation methods designed to capture citizens’ uncertain expectations. Our results suggest that an elicitation method originally proposed by Manski (2009) performs well. It measures average citizens’ subjective belief distributions reliably and is easily implemented in the context of regular (online) surveys. We expect that a wider use of this method will lead to considerable improvements in the study of citizens’ expectations and beliefs.

Although previous scholars have used image data to answer important political science questions, less attention has been paid to video-based measures. In this study, I use motion detection to understand the extent to which members of Congress (MCs) literally cross the aisle, but motion detection can be used to study a wide range of political phenomena, like protests, political speeches, campaign events, or oral arguments. I find not only are Democrats and Republicans less willing to literally cross the aisle, but this behavior is also predictive of future party voting, even when previous party voting is included as a control. However, this is one of the many ways motion detection can be used by social scientists. In this way, the present study is not the end, but the beginning of an important new line of research in which video data is more actively used in social science research.

The two-way linear fixed effects regression (2FE) has become a default method for estimating causal effects from panel data. Many applied researchers use the 2FE estimator to adjust for unobserved unit-specific and time-specific confounders at the same time. Unfortunately, we demonstrate that the ability of the 2FE model to simultaneously adjust for these two types of unobserved confounders critically relies upon the assumption of linear additive effects. Another common justification for the use of the 2FE estimator is based on its equivalence to the difference-in-differences estimator under the simplest setting with two groups and two time periods. We show that this equivalence does not hold under more general settings commonly encountered in applied research. Instead, we prove that the multi-period difference-in-differences estimator is equivalent to the weighted 2FE estimator with some observations having negative weights. These analytical results imply that in contrast to the popular belief, the 2FE estimator does not represent a design-based, nonparametric estimation strategy for causal inference. Instead, its validity fundamentally rests on the modeling assumptions.

This letter reports the results of a study that combined a unique natural experiment and a local randomization regression discontinuity approach to estimate the effect of polls on turnout intention. We found that the release of a poll increases turnout intention by 5%. This effect is robust to a number of falsification tests of predetermined covariates, placebo outcomes, and changes in the time window selected to estimate the effect. The letter discusses the advantages of the local randomization approach over the standard continuity-based design to study important cases in political science where the running variable is discrete; a method that may expand the range of empirical topics that can be analyzed using regression discontinuity methods.

The prespecification of the network is one of the biggest hurdles for applied researchers in undertaking spatial analysis. In this letter, we demonstrate two results. First, we derive bounds for the bias in nonspatial models with omitted spatially-lagged predictors or outcomes. These bias expressions can be obtained without prior knowledge of the network, and are more informative than familiar omitted variable bias formulas. Second, we derive bounds for the bias in spatial econometric models with nondifferential error in the specification of the weights matrix. Under these conditions, we demonstrate that an omitted spatial input is the limit condition of including a misspecificed spatial weights matrix. Simulated experiments further demonstrate that spatial models with a misspecified weights matrix weakly dominate nonspatial models. Our results imply that, where cross-sectional dependence is presumed, researchers should pursue spatial analysis even with limited information on network ties.

Multinomial logit (MNL) differs from many other econometric methods because it estimates the effects of variables upon nominal, not ordered outcomes. One consequence of this is that the estimated coefficients vary depending upon a researcher’s decision about the choice of a reference, or “baseline,” outcome. Most researchers realize this in principle, but many focus upon the statistical significance of MNL coefficients for inference in the same way that they use the coefficients from models with ordered dependent variables. In some instances, this leads researchers to report statistics that do not reflect the correct quantities of interest and reach flawed conclusions. In this note, I argue that researchers need to orient their approach to analyzing both the substantive and statistical significance of predicted probabilities of interest that match their research questions.

Debate on the use of lagged dependent variables has a long history in political science. The latest contribution to this discussion is Wilkins (2018, Political Science Research and Methods, 6, 393–411), which advocates the use of an ADL(2,1) model when there is serial dependence in the outcome and disturbance. While this specification does offer some insurance against serially correlated disturbances, this is never the best (linear unbiased estimator) approach and should not be pursued as a general strategy. First, this strategy is only appropriate when the data-generating process (DGP) actually implies a more parsimonious model. Second, when this is not the DGP—e.g., lags of the predictors have independent effects—this strategy mischaracterizes the dynamic process. We clarify this issue and detail a Wald test that can be used to evaluate the appropriateness of the Wilkins approach. In general, we argue that researchers need to always: (i) ensure models are dynamically complete and (ii) test whether more restrictive models are appropriate.

Binary outcome models are frequently used in the social sciences and economics. However, such models are difficult to estimate with interdependent data structures, including spatial, temporal, and spatio-temporal autocorrelation because jointly determined error terms in the reduced-form specification are generally analytically intractable. To deal with this problem, simulation-based approaches have been proposed. However, these approaches (i) are computationally intensive and impractical for sizable datasets commonly used in contemporary research, and (ii) rarely address temporal interdependence. As a way forward, we demonstrate how to reduce the computational burden significantly by (i) introducing analytically-tractable pseudo maximum likelihood estimators for latent binary choice models that exhibit interdependence across space and time and by (ii) proposing an implementation strategy that increases computational efficiency considerably. Monte Carlo experiments show that our estimators recover the parameter values as good as commonly used estimation alternatives and require only a fraction of the computational cost.