Bridging the Grade Gap: Reducing Assessment Bias in a Multi-Grader Class (with Sean Kates, Tine Paulsen and Joshua Tucker)
Political Analysis (2023)
Replication Archive | R package: [Github]

Many large survey courses rely on multiple professors or teaching assistants to judge student responses to open-ended questions. Even following best practices, multiple graders can result in students with similar levels of conceptual understanding to receive widely varying assessments. We detail how this can occur, and argue that it is an example of differential item functioning (or interpersonal incomparability), where graders interpret the same possible grading range differently. Using both actual assessment data from a large survey course in Comparative Politics and simulation methods, we show that the bias can be corrected for by a small number of ``bridging'' observations across graders. We conclude by offering best practices for fair assessment in large survey courses.

Incentive Effects of Recall Elections: Evidence from California Superior Court Judges (with Sanford Gordon)
Journal of Politics (2022)
Supplementary Appendix | Replication Archive | Media: The Recall, Reframed | Ungated Version

39 U.S. states authorize recall elections, but the incentives they create are understudied. We examine how changes in the perceived threat of recall alter behavior of one set of officials: judges. In 2016, outrage over the sentence imposed on a Stanford athlete following his sexual assault conviction sparked a drive to recall the presiding judge. Using case disposition data from six California counties and matched arrest records for a subset of defendants, we examine whether critical events in the recall campaign were accompanied by corresponding changes in other judges' sentences. We find a large, discontinuous increase in severity associated with the campaign's announcement, but not the recall itself -- suggesting the announcement shifted judges' beliefs about their political environment. The increase may have indirectly burdened minority defendants disproportionately. Our findings are the first to document incentive effects of recall, and suggest that targeted political campaigns may have far-reaching, unintended consequences.

Judicial Accountability and Racial Disparity in Criminal Appeals (with Anna Harvey)
Journal of Legal Studies (2021)
Supplementary Appendix | Replication Archive (Zip file) | Ungated Version
Existing work indicates that retention through election induces larger effects on judicial votes in criminal cases than retention through appointment. Yet existing work has addressed neither case selection effects across retention institutions nor heterogeneous treatment effects by defendant and judge race. Leveraging the unique retention institutions governing New York State's intermediate appellate judges, we report the first within-justice estimates of the effects of reelection and reappointment incentives on judicial votes in criminal appeals. We find that impending judicial reappointment induces a 49 - 52% within-justice decrease in pro-defendant votes in appeals involving Black defendants heard by all-white panels but does not affect votes in other cases. We find no additional effect of impending reelection on appellate justice votes in criminal appeals. Our findings suggest the need for greater attention devoted both to potential selection effects and to heterogeneous effects by defendant and judge race in studies of judicial retention institutions.

Sustaining Democracy with Force: Evidence from Black Enfranchisement during Reconstruction (with Mario Chacon and Jeff Jensen)
Journal of Historical Political Economy (2021)
Winner of 2022 Lee J. Alston Prize | Broadstreet Blog | Ungated Version

The failure of Reconstruction is widely seen as a key factor in the social and economic status of African Americans today. Despite the extension of the franchise to the formerly enslaved, Southern elites used violence and other extralegal means to regain power and ultimately remove these newly granted rights. In this paper, we study the importance of enforcement of political rights on the ability of the formerly enslaved to achieve political power during Reconstruction. We use data on the location of federal troops to predict the election of black politicians in the Congressionally-mandated state constitutional conventions and subsequent state legislatures. Using various estimation strategies, we find that the federal enforcement enhanced black representation and that the presence of the Army interacted positively with other federal efforts such as the Freedmen’s Bureau. In light of the recent Supreme Court decisions to weaken the enforcement mechanisms of the Voting Rights Act and subsequent legislative efforts to suppress minority turnout, our evidence has implications on minority representation to this today.

Democratic Reversals and the Size of Government (with Jeff Jensen)
European Journal of Political Economy (2019)
Ungated Version

While the fiscal and redistributive consequences of democracy is one of the central debates in political economy, most empirical studies analyze this question solely in the context of transitions to democracy. In this paper, we explore the consequences to taxation of democratic reversal using the systematic disenfranchisement of African Americans in the US South between 1880 and 1910. Following the federally-imposed extension of the franchise to the former slaves during Reconstruction (1865–1877), Southern states erected a series of legal restrictions, such as literacy tests and poll taxes, aimed primarily at preventing Southern African Americans from registering to vote. Using an original dataset of local and state taxes and a difference-in-differences estimation strategy, we demonstrate that the adoption of literacy tests for voting eligibility in each state was followed by a significant decline in tax revenues that is highly correlated to the share of each county's population who was African American. We also find that black disenfranchisement led to a shift of the tax burden onto urban counties and a greater reliance on indirect taxation. Our results survive a battery of robustness checks, alternative specifications and additional tests of the redistributionist thesis. The findings are not only consistent with standard models of redistribution following democratization, but also indicate that the elasticity of taxes with respect to enfranchisement is substantial and larger than the one suggested by the cross-national literature.

Under Review

Does Prosecutor Partisanship Exacerbate the Racial Charging Gap? Evidence from District Attorneys in Three States

In the United States, nearly all state prosecutors are elected. The policy discretion afforded to these officials has raised concerns that they may exercise their coercive authority in ways that exacerbate racial disparities. To what extent do local prosecutors' political preferences and electoral incentives affect the charges they bring against defendants from different racial groups? Using linked criminal records from three large states, I find that marginally elected Republican prosecutors seek significantly tougher charges than Democrats but only in cases involving Black defendants. Additional tests demonstrate that observable defendant characteristics (including sex, prior criminal history, crime type, and arrest offense) and sample selection bias (stemming from police expectations of prosecutor punitiveness) cannot explain this result. Further tests indicate that prosecutor partisanship does not merely affect charging outcomes but also shapes disparities in incarceration lengths. The final part of this article examines the extent to which electoral selection or electoral incentives drive these overall effects. One incentive-based explanation predicts differences in reelection concerns across parties. However, the racial gap is largely eliminated in election years as Republican prosecutors get tough all around. An alternative explanation predicts differences in local electoral coalitions across parties due to residential segregation. Consistent with this mechanism, Black defendants tend to reside in precincts where Republican candidates perform poorly.

Base Rate Neglect and the Diagnosis of Partisan Gerrymanders (with Sanford Gordon)
We discuss conceptual issues arising from base rate neglect in the ``Efficiency Gap'' (Stephanopolous and McGhee 2015), a popular measure of unfairness in partisan gerrymandering. A measure of vote dilution asymmetry that adjusts for the base rate mitigates these; however, because it prioritizes a potentially unachievable degree of proportionality, its naive application may be impractical or infeasible. A comparison of measures using enacted and simulated plans clarifies the frequency with which the base rate issue may yield erroneous inferences if unaddressed. We show that a generalized version of dilution asymmetry that differentially weights harms associated with partisan gerrymandering permits and may even necessitate departures from pure proportionality. We also show via simulation that maps that approximate proportionality, and thus minimize dilution asymmetry, are feasible more often than previously thought.

Working Papers

Partisanship, Political Geography and Criminal Sentencing in Texas

Eleven states authorize partisan elections for trial judges. Does judge party influence their sentencing decisions? If so, to what extent do political differences and electoral incentives amplify or mitigate racial sentencing disparities? Despite the near consensus that judges' discretion in this arena is considerable, evidence around the influence of political affiliation on judges' use of discretion has been scant and inconclusive. This study combines millions of criminal records to isolate the causal impact of elected judges' party affiliation on criminal sentencing. I find that Republican judges issue more convictions than Democratic ones presiding in the same county and that this disparity is increasing in the Democratic partisanship of the defendant's precinct. Additional tests indicate that this effect is not driven by selection into arrest or indictment, persistent neighborhood factors, or defendant characteristics. The overall pattern is partly attributable to shifts in the composition of convictions induced by changes in pretrial judicial conduct. Because of the correlation between neighborhood partisanship and racial composition, this pattern may contribute to widespread racial disparities in criminal justice outcomes.