Representation, Elections and Campaigns
"Mediating the Electoral Connection: The Information Effects of Voter Signals on Legislative Behavior" (with John Brooks). (Online Appendix), Journal of Politics 78:3, July 2016.
"Cause or Effect? Turnout in Hispanic Majority-Minority Districts" (with Jasjeet Sekhon and Rocio Titiunik). (Online Appendix), Political Analysis 24:3 Summer 2016.
Legislative redistricting alters the political and electoral context for some voters but not others, thus offering a potentially promising research design to study many questions of interest in political science. We apply this design to study the effect that descriptive representation has on co-ethnic political engagement, focusing on Hispanic participation following California’s 2000 redistricting cycle. We show that when redistrictors draw legislative boundaries in California’s 1990, 2000 and 2010 apportionment cycles, they systematically sort higher participating Hispanic voters into majority-Hispanic (MH) jurisdictions represented by co-ethnic candidates, biasing subsequent comparisons of Hispanic participation across districts. Similar sorting occurs during redistricting in Florida and Texas, though here the pattern is reversed, with less participating Hispanic voters redistricted to MH districts. Our study highlights important heterogeneity in redistricting largely unknown or under-appreciated in previous research. Ignoring this selection problem could significantly bias estimates of the effect of Hispanic representation, either positively or negatively. After we correct for these biases using a hierarchical genetic matching algorithm, we find that, in California, being moved to a district with an Hispanic incumbent has little impact on Hispanic participation in our data.
We utilize a novel experimental design to assess voter selectivity to political advertising. We randomly expose respondents to comparable positive or negative ads aired by Democratic or Republican candidates from the 2012 Presidential race and the 2013 Virginia Gubernatorial contest. The experiment closely mirrors real consumption of campaign information by allowing subjects to skip ads after five seconds, re-watch and share ads with friends. Using these measures of ad-seeking behavior, we find little evidence that negativity influences self-exposure to election advertising. We find partisans disproportionately tune out ads aired by their party’s opponents, though this behavior is asymmetric: Republican-identifiers are more consistent screeners of partisan ads than Democrats. The results advance our understanding of selectivity, showing that party source, and not ad tone, interacts with partisanship to mediate campaign exposure. The findings have important implications about the role self-exposure to information plays in campaigns and elections in a post-broadcast era.
Participation and Partisanship
"Who Matches? Propensity Scores and Bias in the Causal Effects of Education on Participation" (with Sara Chatfield), Journal of Politics 73:3, July 2011.
In a recent study, Kam and Palmer (2008) employ propensity score matching to assess whether college attendance causes participation after reducing selection bias due to pre-adult factors. After matching the authors find no correlation, upending a major pillar in political science. However, we argue that this study has serious flaws and should not be the basis for rejecting the traditional view of an "education effect" on participation. We match on 766,642 propensity scores and use genetic matching to recover better matches with lower covariate imbalances. We consistently find positive effects as covariate balance improves, though no matching approach yields unbiased results. We demonstrate that selection is a serious concern in studying the participatory effects of college attendance and that balance in the covariates and robustness to sensitivity diagnostics should be the ultimate guide for conducting matching analyses.
"Untangling the Education Effect: Moving Educational Interventions into the Experimental Frontier" (with Sara Chatfield), Resources, Engagement, and Recruitment: New Advances in the Study of Civic Volunteerism, Casey A. Klofstad (Ed.), Temple University Press, 2016.
Untangling the precise education effect has proven to be challenging. Directly manipulating educational outcomes through experimentation is usually infeasible and unethical. And there is a serious concern that non-random pressures to seek out more education remain problematic in observational studies, and perhaps even those utilizing natural experiments. Our aim in this chapter is to outline a roadmap for future research on the political returns to schooling, given the limits to experimentation in educational attainment. We think developing such a guide will be helpful for political scientists as the field moves increasingly into the experimental frontier. This is especially so given a possible future where scholars have exhausted new sources of exogenous natural variation, are unable to experimentally manipulate years of education, and yet remain skeptical of much of the observational findings about education.
Educational attainment is robustly associated with greater political participation, yet the causal nature of this finding remains contested. To assess this relationship, I leverage a natural experiment in the Rockefeller Sanitary Commission's (RSC) anti-hookworm campaign, which exogenously expanded primary and secondary education in the early-20th century American South. I evaluate two RSC hookworm interventions: exposure to the campaign and proportion treated. I use genetic matching to control for observable factors that influenced the haphazard dispensing of treatment, and develop new matching methods for continuous campaign interventions. I also use a variety of methods to assess the robustness of the results to a number of alternative accounts. Throughout, I find a consistent positive effect of education on participation, suggesting additional evidence for a causal interpretation of the ‘education effect’.
Political parties can provide valuable information to voters by cultivating distinct associations between their labels, issue priorities, policies and group traits. Yet, there is considerable debate over which associations voters incorporate, and whether these are accurate. In this study, we develop a novel conjoint classification experiment designed to map voters’ partisan associative networks. We ask respondents to ‘guess’ the party and ideology of hypothetical candidates given randomized issue priorities and biographical details. Notably, this inferential approach minimizes the biasing effects of partisan boosting in measuring the relative associations voters make between attributes and parties, and the impact these mappings have on candidate evaluations. We find voters consistently link many issues with party and ideological labels, but agree far less on associations with candidate attributes. Our study highlights important heterogeneity in the information value of party reputations, with implications for theories of democratic competence and empirical findings emerging from candidate-vignette designs.
A generation of men were randomly assigned lottery numbers to determine draft status during the Vietnam War, offering researchers a natural experiment to assess the behavioral effects of differential exposure to the risk of military service. However, conventional analysis of these effects may be biased by out-of-sample attrition influenced by draft risk. Additionally, the effects of the lottery may be unconventional or heterogeneous - low numbers could elicit an 'empathetic' effect on draft-age men ineligible due to prior military service. In this study, I use the Youth Parent Socialization Panel Study (YPSPS) data to assess whether the Vietnam draft lottery influenced attrition behavior in later waves of the survey, and whether these and other effects depend heterogeneously on pre-1969 military service. I find that high draft risk decreases attrition for draft-eligible men, but has no impact on the ineligible. I then impute missing data over multiple waves to reevaluate the impact of the lottery on participation and attitudes accounting for sample attrition, and validate these imputations through a variety of tests. Overall, I find that either the behavioral effects of the draft uncovered in prior studies are substantially overstated, or that the mere assignment of low draft numbers demobilized men insulated from draft risk, and pushed their attitudes and partisanship in a conservative direction.
Causal Inference and Estimation
"Maximum Entropy Imputation".
In this study, I develop a novel methodological approach, maximum entropy imputation (MEI), to correct for attrition bias when estimating causal effects in experiments. MEI reweights observations to construct a synthetic sample from the group of non-attritioners that near-perfectly resemble those who actually attrition in their observable characteristics. I introduce and discuss the assumptions of MEI for identifying causal effects with sample attrition. I then present the results from a series of validation tests and simulations that assess the performance of the imputation approach. Finally, I employ MEI to replicate a number of behavioral findings in political science and epidemiology.
"A Genetic Matching Approach to Estimating Treatment Effects Using Non-Binary Interventions".
I develop a new evolutionary approach to maximizing covariate balance when matching using continuous or non-binary interventions. The aim in the approach is to match units to minimize distances on covariates, while maximizing differences on the continuous treatments, so that there are ’high’ and ’low’ dose units in each matched pair. Balance is maximized on covariates across the high and low doses, us- ing weights to heighten differences on some covariates over others during matching. An evolutionary algorithm is used to identify the weights that optimally minimize covariate differences over successive generations. Through simulations and empirical applications, I show that continuous genetic matching generally performs much better that alternative parametric and matching estimation approaches.
Scaling Text and Machine LearningThough powerful as general tools, automated measures of position-taking in text often perform poorly when models of speech are difficult to develop or theoretically contested. Rather than model text, I develop an experimental approach to measure perceptions of partisanship in speech, with an application to 2008 Congressional advertisements. I randomly assign ads to subjects recruited in a large-N survey, and ask them to 'guess' the party of featured candidates, with ads scored as their average party inference. These party perception scores are empirically synonymous with a liberal-conservative dimension, and highly reliable across samples and experimental conditions. Party identity has little impact on guesses, indicating the inferential task significantly mutes partisan bias. For validation, I assess which words influence guessing, and whether ad-scores correspond to expectations about how candidates target voters. Importantly, this experimental approach can augment or validate automated text analysis, and generalize to study speech across many other contexts.