Current Research

I have two main research interests with American Politics.  The first, within judicial politics, is how collegiality and hierarchy shape and constrain decision making in the U.S. Supreme Court and lower courts.  The second, within distributive politics, is the study of the ways in which both public opinion and U.S. institutions, like parties and apportionment schemes, leave their mark on distributive spending in states and localities.

I am also interested in studying techniques for modeling and analyzing clustered (grouped) data, as many of the causal forces at the core of my substantive inquiries take the form of group-level treatments.

Judicial Politics

Circuit Splits on the U.S. Courts of Appeals. Read more about this project and our data here.

Legal Uniformity in American Courts with Deborah Beim. 

Abstract: Intercircuit splits occur when two or more circuits on the U.S. Courts of Appeals issue different legal rules about the same legal question. When this happens, federal law is applied differently in different parts of the country. Intercircuit splits cause legal non-uniformity, are an impediment to lawyering and judging, and have practical consequences for American law. Despite intercircuit splits’ importance, there is almost no quantitative research about them. We created a unique original dataset that includes intercircuit splits that arose between 2005 and 2013. For each intercircuit split, we identified every circuit and every case involved. These data reveal that one-third of intercircuit splits are resolved by the Supreme Court. Two-thirds are not. We show that those that will be resolved are resolved within three years after they arise, and we show that splits are more likely to be resolved when they exhibit contemporaneous and growing disagreement. However, many such splits are never resolved by the Supreme Court. Those that are not resolved by the Supreme Court continue to yield litigation and do not dissipate on their own, and the likelihood of resolution does not rise as time passes.

Americans’ Knowledge of the U.S. Supreme Court with John Bullock.

Abstract: Recent and influential research suggests that political scientists have sharply understated popular knowledge of the U.S. Supreme Court. This research implies that popular knowledge of other aspects of government has been understated as well. The implication may be correct. But these revisionist studies of political knowledge err in the other direction: they overstate popular knowledge of politics. Focusing on the Supreme Court, we use a series of national-sample experiments to show that inferences about popular knowledge of politics depend heavily on little-appreciated aspects of survey design and analysis. Accounting for these aspects of design and analysis suggests a level of knowledge in the polity that lies between the levels suggested by conventional and revisionist research.

Distributive Politics 

When Nothing is Better Than Something: How Racial Attitudes Shape Public Support for Government Spending, with Katherine Krimmel.

Abstract: We provide and test a theory explaining how and why racial attitudes shape public opinion on government spending in the United States. We hypothesize that many people think the government allocates money unfairly across racial groups, and “inequity aversion” leads them to reject spending as a result. Using data from an original survey, we show many people disapprove of the distribution of government money across races, and these views affect their support for spending. Through a series of survey experiments, we then show that people who think the distribution of spending is unfair to whites will oppose the establishment of otherwise desirable spending programs and even reject outlays therefrom if some of the money might go toward racial minorities. While it is generally assumed that “goodies” are always good from an electoral perspective, nothing is better than something if a portion of that something would benefit minorities.

The Federal Spending Paradox: Economic Self-Interest and Symbolic Racism in Contemporary Fiscal Politics with Katherine Krimmel.  (Appendix) (Historical epilogue)

Abstract: We show how symbolic politics condition public opinion on federal spending, and how this helps to explain an important puzzle in contemporary American politics.  Using multilevel regression and poststratification to estimate state-level opinion on federal spending, we show that, curiously, opposition to federal spending is higher in states receiving more federal money, per tax dollar paid.  Belying the popular narrative surrounding so-called “red state socialism,” we find that simple hypocrisy does not explain this paradox—individuals who are likely to benefit from spending tend to support it.  But, income is a more powerful predictor of opinion on spending in “taker” states than “giver” states, heightening state-level opposition in the former.  There is also more to the story than economic self-interest.  Symbolic racism is four times more powerful than income in explaining opposition to spending, and there are more people with such attitudes in states receiving more federal money.

In the historical epilogue, we add a time dimension to the paradox.  This provides the foundation for a larger project on the changing dynamics of fiscal politics over the postwar period.

How Should We Estimate State-Level Opinion on Quota-Sampled Data from the 1930s and 1940s? An Evaluation of Multilevel Regression and Poststratifi cation with Kate Krimmel.

Party Effects on the Distribution of Federal Outlays: A Regression Discontinuity Approach with Austin Nichols. 

Abstract: Do parties affect the pattern of distributive spending in the U.S. House of Representatives?Party-cartel theories of legislative organization posit that majority party status accords party members influence over legislative outcomes that they would not otherwise wield. The implication for distributive spending is that a majority party member elected from a given district could extract more distributive benefits for the district than a could minority party member even if elected by the same district for the same Congress. We use a regression discontinuity design that exploits a particular feature of the U.S. majoritarian voting scheme—that the majority party status of the winner is “assigned” discontinuously at fifty percent (plus one vote) of the majority party vote share—to investigate the effect of being represented by majority party member on distributive spending in the district. Using data on federal spending in districts from fiscal years 1983 to 2002, we determine if federal spending as a function of the percent of the electorate that voted for the majority party candidate “jumps” at fifty percent. If there is a significant jump in spending near fifty percent, this would indicate that the majority party does indeed use its power to secure more federal funds than the minority party in marginal districts. Our design also allows to test whether Democrats or Republicans are bigger spenders in marginal districts, by exploiting the fact that the party affiliation of a district’s representative is assigned discontinuously at fifty percent (plus one vote) of the two-party vote for the Democrat. Using RD, we find no evidence of a causal effect of majority party status nor of party affiliation on outlays to the district.

Malapportionment in the U.S House of Representatives: The Effect of Census Reapportionment on the Distribution of Federal Funds to States. 


Quantitative Methods and Other Work

Randomization Tests and Inference with Grouped Data

Abstract: Political scientists often ask questions that require making inferences about the effects of variables measured at the group level on outcomes measured at the individual level. Inference with grouped data presents special challenges because the amount of independent information in the data is often more related to the number of groups than to the number of individual observations. A common parametric solution to this problem is to calculate cluster-robust standard errors and to use those errors in hypothesis tests. However, cluster-robust standard errors perform poorly when the number of groups is fewer than 50. I present Monte Carlo evidence which shows that, in terms of Type I error rates, the non-parametric randomization test outperforms t-tests using cluster-robust standard errors regardless of the number of groups. In terms of power, the loss from using randomization tests is small. Thus, randomization tests are a viable alternative to the cluster-robust approach, particularly when the number of groups is small.

Analyzing Democratic Trade: When Less is More, with Robert Erikson and Pablo Pinto

Abstract: Does being a democracy cause a nation to seek out other democracies as trade partners? Research on this question has a long history, whereby the unit of analysis is the dyad-year (e.g., Canada-Belgium, 1994). However, dyadic analysis yields conventional tests of significance that are over-confident. This paper replaces the dyad with the nation as the unit of analysis. It asks whether nations that undergo democratic (or autocratic) transformations shift their trade toward (against) democracies. With this design, we find strong statistical evidence for the democratic trade hypotheses. We suggest that many research questions where dyadic analysis is the norm can more successfully be conducted with a difference-in-differences design.

Election Laws and Voter Turnout Among the Registered: What Causes What? with Robert Erikson. 

Abstract: Scholars make causal claims about the vote-boosting power of laws designed to increase turnout, citing evidence from regression analyses that show that generous voting laws are related to high turnout. Yet one must be skeptical of contamination from endogeneity in this relationship. The skeptic’s argument is: states with a culture of participation pass legislation designed to encourage voting. With participatory states being the cause of pro-turnout legislation, the causal direction is reversed from what is normally supposed. We take the skeptic’s argument seriously and use sensitivity tests to evaluate claims that turnout is influenced by pro-turnout legislation (and vise versa). Specifically, we apply a “zero covariance restrictions” assumption and estimate the effect of turnout on legislation via two-stage least squares, with state demographic variables as instruments. Then, assuming that there are no unobserved variables that affect both turnout and legislation, we can back out the reverse effect of legislation on the vote. The 2SLS analysis shows that the turnout-legislation effect fully accounts for the turnout-legislation covariance, leaving no room for legislation to affect turnout.