The Program for Quantitative and Analytical Political Science (Q-APS)

In late 2004, Ukraine’s “Orange Revolution” captured international attention as an example of citizens seizing control of their country’s future. Recently, though, the seemingly unthinkable happened in Ukraine: Viktor Yanukovich – whose blatant attempts to steal the 2004 Ukrainian presidential elections using all manners of electoral fraud gave rise to the Orange Revolution – was elected president of Ukraine and inaugurated with almost no popular response. These events present a puzzle: why were citizens willing to bear the cost of protesting in 2004 to ensure that Yanukovich not become president, only to shrug their collective shoulders at his victory five years later? We consider this puzzle from a theoretical perspective, modeling the effect of protest at one point in time on the likelihood of protest in the future. The model provides numerous predictions for why we might observe a “one shot deal” scenario whereby protest occurs in the first period but not the second, but the most important novel insight concerns the learning process of citizens. We suggest that citizens may not only be discovering the type of their new government - as most previous models of adverse selection assume - but rather may also be learning about the universe of potential governments in their country. In doing so, we expand the formal literature on the adverse selection problem in selecting governments in two important new directions: conditions under which removing governments is costly, and conditions under which the underlying distribution of possible government types is imperfectly known.

The evolution of cooperation in nature and human societies depends crucially on how the benefits from cooperation are divided and whether individuals have complete information about their payoffs. We tackle these questions by adopting a methodology from economics called mechanism design. Focusing on reproductive skew as a case study, we show that full cooperation may not be achievable due to private information over individuals’ outside options, regardless of the details of the specific biological or social interaction. Further, we consider how the structure of the interaction can evolve to promote the maximum amount of cooperation in the face of the informational constraints. Our results point to a distinct avenue for investigating how cooperation can evolve when the division of benefits is flexible and individuals have private information.

Experimentation is a powerful methodology that enables scientists to empirically establish causal claims. However, one important criticism is that experiments merely provide a black-box view of causality and fail to identify causal mechanisms. Specifically, critics argue that although experiments can identify average causal effects, they cannot explain the process through which such effects come about. If true, this represents a serious limitation of experimentation, especially for social and medical science research that strive to identify causal mechanisms. In this paper, we consider the several different experimental designs that help identify average natural indirect effects. Some of these designs require the direct manipulation of an intermediate variable, while others can be used even when only imperfect manipulation is possible. We use recent social science experiments to illustrate the key ideas that underlie each of the proposed designs.

When interest groups compete to influence legislators, the resulting legislation is often vague, and thus obliges the groups to continue their fight in the executive. On its face, this seems inefficient—at least from the point of view of the groups. We explore this intuition in a model of “nested lobbying” in which interest groups first compete to influence a legislative agenda setter, then compete to influence legislative votes over the resulting agenda. If the resulting legislation grants discretion to the executive, the final prize is allocated in yet one more contest in the bureaucracy. We find that when the status quo is non-discretionary, competition over the agenda never results in an agenda that includes discretion. Surprisingly, however, a discretionary status quo can stand with probability 1 if the preferences of the bureaucracy, the legislature, and the agenda setter are arranged in an “iron triangle”. Specifically, the bureaucracy and agenda setter must be biased in favor of one group, while the legislature is biased in favor of the other.

The validity of empirical research often relies upon the accuracy of self-reported behavior and beliefs. Yet, eliciting truthful answers in surveys is challenging especially when studying sensitive issues such as racial prejudice, corruption, and support for militant groups. List experiments have attracted much attention recently as a potential solution to this measurement problem. Many researchers, however, have used a simple difference-in-means estimator without being able to efficiently examine multivariate relationships between respondents' characteristics and their answers to sensitive items. Moreover, no systematic means exist to investigate role of underlying assumptions. We fill these gaps by developing a set of new statistical methods for list experiments. We identify the commonly invoked assumptions, propose new multivariate regression estimators, and develop methods to detect and adjust for potential violations of key assumptions. For empirical illustrations, we analyze list experiments concerning racial prejudice. Open source software is made available to implement the proposed methodology.

Empirical testing of competing theories lies at the heart of social science research. We demonstrate that a well-known class of statistical models, called finite mixture models, provides an effective way of rival theory testing. In the proposed framework, each observation is assumed to be generated either from a statistical model implied by one of the competing theories or more generally from a weighted combination of multiple statistical models under consideration. Researchers can then estimate the probability that a specific observation is consistent with each rival theory. By modeling this probability with covariates, one can also explore the conditions under which a particular theory applies. We discuss a principled way to identify a list of observations that are statistically significantly consistent with each theory, and propose measures of the overall performance of each competing theory. We illustrate the relative advantages of our method over existing methods through empirical and simulation studies.

It is well known that during a crisis unitary rational states have an incentive to misrepresent their true resolve and willingness to go to war. This theoretical result has been taken to imply that diplomacy, interpreted as pre-bargaining communication, can have no effect on the way crises play out. This paper shows an intuitive way that diplomatic cheap talk can matter in a single crisis between countries, especially when the bargaining game has multiple equilibria. In particular, if after 'diplomacy' states can choose to either fight a war directly or bargain in hopes of reaching a peaceful settlement, then it is possible to find an equilibrium where diplomacy influence whether there is war or peace. Importantly the cheap talk diplomacy does three things the standard model says it cannot: it coordinates actions, it reveals information, and it changes the ex ante probability of war. This result demonstrates an easy way of reconciling the discrepancy between the obvious empirical observation that diplomacy often does influence the path of a crisis and the rationalist model of war.

eco is a publicly available R package that implements the Bayesian and likelihood methods proposed in Imai, Lu, and Strauss (2008b) for ecological inference in 2x2 tables as well as the method of bounds introduced by (Duncan and Davis 1953). The package ts both parametric and nonparametric models using either the Expectation-Maximization algorithms (for likelihood models) or the Markov chain Monte Carlo algorithms (for Bayesian models). For all models, the individual-level data can be directly incorporated into the estimation whenever such data are available. Along with in-sample and out-of-sample predictions, the package also provides a functionality which allows one to quantify the effect of data aggregation on parameter estimation and hypothesis testing under the parametric likelihood models. This paper illustrates the usage of eco with several real data examples that are also part of the package.

Political scientists have recently conducted hundreds of randomized field experiments to examine the effectiveness of various mobilization methods for increasing voter turnout. Given the high degree of internal and external validity, the empirical findings of these studies have a potential to significantly impact the practice of get-out-the-vote (GOTV) campaigns in the real world. In this paper, we offer an essential and yet missing methodological tool that allows GOTV campaign planners to best utilize the results of such field experiments. In particular, we show how to derive the optimal GOTV campaign strategy from field experiments. Our nonparametric method is applicable to partisan or nonpartisan campaigns as well as campaigns with multiple mobilization methods of the same or different costs. We evaluate the effectiveness of the proposed method using three existing field experiments. In multiple cases, we find that the resulting optimal campaign strategy is more than twice as cost-effective as a naive strategy.
This article has won the Political Analysis Editors' Choice Award

MatchIt implements the suggestions of Ho, Imai, King, and Stuart (2007) for improving parametric statistical models by preprocessing data with nonparametric matching methods. MatchIt implements a wide range of sophisticated matching methods, making it possible to greatly reduce the dependence of causal inferences on hard-to-justify, but commonly made, statistical modeling assumptions. The software also easily ts into existing research practices since, after preprocessing data with MatchIt, researchers can use whatever parametric model they would have used without MatchIt, but produce inferences with substantially more robustness and less sensitivity to modeling assumptions. MatchIt is an R program, and also works seamlessly with Zelig.

The item count technique is a survey methodology that is designed to elicit respondents' truthful answers to sensitive questions such as racial prejudice and drug use. The method is also known as the list experiment or the unmatched count technique and is an alternative to the commonly used randomized response method. In this paper, I propose new nonlinear least squares and maximum likelihood estimators for efficient multivariate regression analysis with the item count technique. The two-step estimation procedure and the Expectation Maximization algorithm are developed to facilitate the computation. Enabling multivariate regression analysis is essential because researchers are typically interested in knowing how the probability of answering the sensitive question affirmatively varies as a function of respondents' characteristics. As an empirical illustration, the proposed methodology is applied to the 1991 National Race and Politics survey where the investigators used the item count technique to measure the degree of racial hatred in the United States. Small-scale simulation studies suggest that the maximum likelihood estimator can be substantially more efficient than alternative estimators. Statistical efficiency is an important concern for the item count technique because indirect questioning means loss of information. The software package is made available to implement the proposed methodology.

Fluctuations in the price of oil and the contemporaneous political changes in oil-producing countries have raised an important question about the link between oil rents, political institutions, and civil liberties. This article presents a simple model of the relationship between resource income and political freedom and, using an instrumental variables approach, estimates the causal effect of shocks to oil revenues on levels of democracy. Using a new data set, multiple measures of democracy, and various specifications, I find that the effect of oil price shocks is larger than might be expected and on the order of the effects found from changes in gross domestic product.

Political scientists have long been interested in citizens' support level for socially sensitive actors such as ethnic minorities, militant groups, and authoritarian regimes. Attempts to use direct questioning in surveys, however, have largely yielded unreliable measures of these attitudes as they are contaminated by social desirability bias and high non-response rates. In this paper, we develop a statistical methodology to analyze endorsement experiments, which recently have been proposed as a possible solution to this measurement problem. The commonly used statistical methods are problematic because they cannot properly combine responses across multiple policy questions, the design feature of a typical endorsement experiment. We overcome this limitation by using item response theory to estimate support levels on the same scale as the ideal points of respondents. We also show how to extend our model to incorporate a hierarchical structure of data in order to recoup the loss of statistical efficiency due to indirect questioning. We illustrate the proposed methodology by applying it to measure political support for Islamist militant groups in Pakistan. Simulation studies suggest that the proposed Bayesian model yields estimates with reasonable levels of bias and statistical power. Finally, we offer several practical suggestions for improving the design and analysis of endorsement experiments.

We study two different varieties of uncertainty that countries can face in international crises and establish general results about the relationship between these sources of uncertainty and the possibility of peaceful resolution of conflict. Among our results, we show that under some weak conditions, there is no equilibrium of any crisis bargaining game that has voluntary agreements and zero probability of costly war. We also show that while uncertainty about the other side’s cost of war may be relatively benign in peace negotiations, uncertainty about the other side’s strength in war makes it much more difficult to guarantee peaceful outcomes. Along the way, we are able to assess the degree to which particular modeling assumptions found in the existing literature drive the well-known relationship between uncertainty, the incentive to misrepresent, and costly war. We find that while the theoretical connection between war and uncertainty is quite robust to relaxing many modeling assumptions, whether uncertainty is about costs or the probability of victory remains important.

Identifying causal mechanisms is a fundamental goal of social science. Researchers seek to study not only whether one variable affects another but also how such a causal relationship arises. Yet, commonly used statistical methods for identifying causal mechanisms rely upon untestable assumptions and are often inappropriate even under those assumptions. Randomizing treatment and intermediate variables is also insufficient. Despite these difficulties, study of causal mechanisms is too important to abandon. We make three contributions to improve research on causal mechanisms. First, we present a minimum set of assumptions required under standard designs of experimental and observational studies and develop a general algorithm for estimating causal mediation effects. Second, we provide a method to assess sensitivity of conclusions to potential violations of a key assumption. Third, we offer alternative research designs for identifying causal mechanisms under weaker assumptions. The proposed approach is illustrated using media framing experiments and incumbency advantage studies.

Political scientists have long been concerned about the validity of survey measurements. Although many have studied classical measurement error in linear regression models where the error is assumed to arise completely at random, in a number of situations the error may be correlated with the outcome. We analyze the impact of differential measurement error on causal estimation. The proposed nonparametric identification analysis avoids arbitrary modeling decisions and formally characterizes the roles of different assumptions. We show the serious consequences of differential misclassification and offer a new sensitivity analysis that allows researchers to evaluate the robustness of their conclusions. Our methods are motivated by a field experiment on democratic deliberations, in which one set of estimates potentially suffers from differential misclassification. We show that an analysis ignoring differential measurement error may considerably overestimate the causal effects. This finding contrasts with the case of classical measurement error which always yields attenuation bias.

Traditionally in the social sciences, causal mediation analysis has been formulated, understood, and implemented within the framework of linear structural equation models. We argue and demonstrate that this is problematic for three reasons; the lack of a general definition of causal mediation effects independent of a particular statistical model, the inability to specify the key identification assumption, and the difficulty of extending the framework to nonlinear models. In this paper, we propose an alternative approach that overcomes these limitations. Our approach is general because it offers the definition, identification, estimation, and sensitivity analysis of causal mediation effects without reference to any specific statistical model. Further, our approach explicitly links these four elements closely together within a single framework. As a result, the proposed framework can accommodate linear and nonlinear relationships, parametric and nonparametric models, continuous and discrete mediators, and various types of outcome variables. The general definition and identification result also allow us to develop sensitivity analysis in the context of commonly used models, which enables applied researchers to formally assess the robustness of their empirical conclusions to violations of the key assumption. We illustrate our approach by applying it to the Job Search Intervention Study (JOBS II). We also offer easy-to-use software that implements all of our proposed methods.

Causal mediation analysis is routinely conducted by applied researchers in a variety of disciplines. The goal of such an analysis is to investigate alternative causal mechanisms by examining the roles of intermediate variables that lie in the causal paths between the treatment and outcome variables. In this paper, we first prove that under a particular version of sequential ignorability assumption, the average causal mediation effect (ACME) is nonparametrically identified. We compare our identification assumption with those proposed in the literature. Some practical implications of our identification result are also discussed. In particular, the popular estimator based on the linear structural equation model (LSEM) can be interpreted as an ACME estimator once additional parametric assumptions are made. We show that these assumptions can easily be relaxed within and outside of the LSEM framework and propose simple nonparametric estimation strategies. Second, and perhaps most importantly, we propose a new sensitivity analysis that can be easily implemented by applied researchers within the LSEM framework. Like the existing identifying assumptions, the proposed sequential ignorability assumption may be too strong in many applied settings. Thus, sensitivity analysis is essential in order to examine the robustness of empirical findings to the possible existence of an unmeasured confounder. Finally, we apply the proposed methods to a randomized experiment from political psychology. We also make easy-to-use software available to implement the proposed methods.

The authors develop a model of alliances with outside options to study burden sharing in nonbinding alliance agreements. The analysis provides an explanation for the variation in ally contributions to NATO over time and why the post–Cold War period has seen an increase in the use of coalitions of the willing. Additionally, the analysis reveals something of an initiator’s disadvantage in burden sharing—the initiator of an alliance action pays a disproportionate cost of the military burden. The authors’ argument provides an alternative explanation for why the United States has been consistently the largest contributor to NATO.

A basic feature of many field experiments is that investigators are only able to randomize clusters of individuals - such as households, communities, firms, medical practices, schools, or classrooms - even when the individual is the unit of interest. To recoup the resulting efficiency loss, some studies pair similar clusters and randomize treatment within pairs. However, many other studies avoid pairing, in part because of claims in the literature, echoed by clinical trials standards organizations, that this matched-pair, cluster-randomization design has serious problems. We argue that all such claims are unfounded. We also prove that the estimator recommended for this design in the literature is unbiased only in situations when matching is unnecessary; and its standard error is also invalid. To overcome this problem without modeling assumptions, we develop a simple design-based estimator with much improved statistical properties. We also propose a model-based approach that includes some of the benefits of our design-based estimator as well as the estimator in the literature. Our methods also address individual-level noncompliance, which is common in applications but not allowed for in most existing methods. We show that from the perspective of bias, efficiency, power, robustness, or research costs, and in large or small samples, pairing should be used in cluster-randomized experiments whenever feasible; failing to do so is equivalent to discarding a considerable fraction of one's data. We develop these techniques in the context of a randomized evaluation we are conducting of the Mexican Universal Health Insurance Program.

We contend that, with a suitably broad notion of rationality and a diverse set of motivations, the game-theoretic tradition is particularly well suited for generating insights about effects of deliberative institutions and that progress in the development of deliberative democratic theory hinges on making proper sense of the relationship between game-theoretic and normative theoretic approaches to deliberation. To advance this view, we explore the central methodological issues at the core of that relationship and address the arguments raised against the relevance of game-theoretic work on deliberation. We develop a framework for thinking about the differences in how the normative and the game-theoretic approaches frame and answer questions about deliberation and articulate an approach to a deliberative democratic theory that builds on the strengths of both of these theoretic traditions, properly informed by empirical scholarship.

In this paper, we consider the possibility of identifying peaceful mechanisms such as bargaining protocols, international institutions, or norms that can enable countries to settle disputes in the absence of binding contracts. In particular, we are interested in the existence of mechanisms with zero probability of war. Here, we focus on situations where the countries’ payoffs to war are interdependent or correlated and where efficient settlements are not required but subsidies are unavailable. Most importantly, countries can choose to go to war at any time and can use information learned from the negotiation process in making this choice. We characterize the conditions under which no peaceful mechanisms exist and discuss how weakening our war consistency condition can change this result.

We consider a two-period model of elections in which voters have private information about their policy preferences. A first-period vote can have two types of consequences: it may be pivotal in the first election and it provides a signal that affects candidates' positions in the second election. Pivot events are exceedingly unlikely, but when they occur the effect of a single vote is enormous. In contrast, vote totals always have some signaling effect, but the effect of a single vote is small. We investigate which effect – pivot or signaling – drives equilibrium voting behavior in large electorates.