Publications

Under Review

Research Interests (Research statement in pdf format)

American Politics

The substantive topic I study is the relationship between elites and the mass public. To investigate this subject, my research agenda focuses on two questions: First, how do elites behave in a broader electoral context? For example, how do they behave if they want to win election for policymaking reasons or if they face an electorate with a changing distribution of ideology? Second, how does elite behavior influence the electorate? I plan to ask these questions for a variety of issues, starting first by studying immigration in my dissertation and moving on to issues like healthcare afterward.

My dissertation, "The Long-Term Consequences of Immigration Politics," investigates these questions about the elite-mass relationship by focusing on an environment in which the issue preferences of the public shift over time with demographic changes. The formal element of this dissertation asks how parties will place themselves on an issue if they know that the ideological distribution of voters will be different in the future than at present. The model reveals that even when politicians are concerned with winning elections not just at present, but also in the future, they still will focus mainly on winning today. Their willingness to make concessions to the electorate of the future, however, increases the more they value winning a future election. The empirical studies focus on the flip side of the relationship by investigating the consequences of elite behavior for the electorate. One study compares the states of California, Texas, & Ohio, and it shows how differences in parties' positions of immigration influence the long-term party affiliations of the state electorate. Another study asks when state legislators will raise the issue of immigration and enact laws on it, thereby revealing the policy consequences of strategic behavior on the part of politicians. Together these three studies present a picture of electoral influences on politicians and politicians' influences on the mass public regarding the topic of immigration.

For my first post-dissertation project, I will extend the two broad questions of my agenda to the issue of healthcare. As an issue, health care offers a chance to understand what policy-motivated elites will do to simultaneously retain their positions and enact their preferences. Healthcare is a relatively complex issue, so elected officials have more choices in how they rhetorically define the issue: they can use a problem-based or a symbolic approach. In these studies, I will argue that those seeking to build a large coalition to enact new policy will be likely to discuss the issue in technical, wonkish terms, while those opposing new legislation will reduce the issue to easy cues. Turning to the consequences of elite behavior for the electorate, this research will contend that voters will be more likely to polarize on the issue by party when public discourse is dominated by cues and symbolism that readily capture their attention. When discourse is policy-oriented, it will be more time consuming to parse out the position a voter feels is consistent with his or her own dispositions. Unwilling to do this, most uninterested citizens will remain indifferent or weakly favorable to new policy proposals, but there will be less of a partisan difference. Hence, how policy can be made and what it means for electoral polarization should be contingent largely on the rhetorical strategies of politicians.

Much of my research with Virginia Gray & David Lowery also has spoken to the consequences of electoral considerations to policymaking. For example, our paper "Public Opinion, Organized Interests, and Policy Congruence in Initiative and Noninitiative States" replicates past research on congruence between public opinion & policy in the states and shows that the presence of ballot initiatives does not condition this relationship. Our paper "Incrementing Toward Nowhere" models legislative action on universal health coverage in the states based on a number of factors, including public opinion and organized interests. These two papers also have been helpful in my dissertation, where I model state policy activity on immigration: a clearer understanding of policymaking has helped me form my theory and empirical model for immigration policy. Lastly, our article in the Journal of Politics, entitled "The Construction of Interest Communities," focuses on the causal process behind the size of interest communities and observes that empirically distinguishing between Olson's theory and Gray and Lowery's can be difficult. All of these studies have contributed to my goal of understanding how the public influences policymakers either with a direct test, with a description of how other factors influence elites, or with a deep explanation of an important factor like interest groups.

Political Methodology

Much of my research in political methodology asks what techniques will yield the best inferences when facing the problems of real data analysis. For instance, my paper from Michigan's 2006 EITM institute, asks what technique of estimating quantal response models can best handle model misspecification. This draws on the research of Curt Signorino, who studies international crisis bargaining games. By assuming that nation-states earn utility from war or peace based on measurable variables, but occasionally make errors in their decisions, he shows how to derive a multinomial probit model directly from the formal model. My paper runs Monte Carlo simulations on this model when estimated by either a single maximum likelihood estimator or sequential probits and finds that estimates from a sequence of probits actually handles model misspecification better than the system model. Since both methods do equally well in an exact specification, this study suggests that researchers using this technique would be safer with a sequence of probits.

As another example of studying robust estimators, additional research from my master's thesis considered a heteroscedasticity hypothesis with an ordinal dependent variable. However, heteroscedastic ordered probit models are known to have a fragile identification in maximum likelihood estimation, and I was working with panel data. To obtain more stable and trustworthy results, then, I programmed a hierarchical Bayesian estimator. This flexible technique allowed me to account for the panel data, an ordinal dependent variable, and nonconstant variance without some of the problems maximum likelihood faces.

Finally, I am committed to taking the best tools available to social science and finding new, relevant uses for them. One example of this comes in a dissertation chapter, "Strategic Party Placement with a Dynamic Electorate," where I use R to program an equilibrium search of an insoluble game. This provided a sense of how the results changed as model parameters varied, even without an analytic solution.

For more information, please download my CV.