The Increasingly United States: How and Why American Political Behavior Nationalized. 2018. Chicago, IL: University of Chicago Press. [Appendix]
[Media coverage: New Yorker, New York Times, FiveThirtyEight, Vox, National Review, Atlantic, Bloomberg, City Lab, Vox: Mischiefs of Faction, Guardian, Harvard Law Review Blog, Behavioral Scientist, Lancaster Online, Honolulu Civil Beat, WGDR’s “Relocalizing Vermont”, Citizen Times, New Haven Register][Podcasts: NPR: It’s Been a Minute, New Books Network, Niskanen Center’s Political Research Digest, Politics Politics Politics] [Academic Reviews: Thomas K. Ogorzalek/Perspectives on Politics, Robert Shapiro/H-DIPLO, Braford Bishop/Publius: The Journal of Federalism, Jamie L. Carson/Presidential Studies Quarterly]
Studies in American Political Development, forthcoming; co-authored with Eric Schickler and David Azizi
We use manual and automated analyses of 1,783 state party platforms to trace the changing issue agenda across American states over a century–and to chart rising polarization and nationalization in recent decades. [Appendix] [Data] [Ungated]
Race, Ethnicity, and Immigration
Journal of Politics, conditionally accepted. Co-authored with Cheryl Kaiser and Efrén Pérez. SSRN Working Paper, 3840922.
We use one of the first-ever panels of Asian American and Latino respondents to trace their partisan identities and related political attitudes and behaviors.
Political Behavior, 2021, 43:(663-686); SSRN Working Paper 3186800
Using a 12-wave panel of the U.S. population conducted between 2007 and 2016, this paper demonstrates that prejudice against Blacks was a strong predictor of Whites’ shifting support for presidential candidates between 2012 and 2016. [Appendix] [SSRN] [Data] [Read Online] [Media: FiveThirtyEight, New York Times]
Public Opinion Quarterly, 2020, 84(1):119-140; SSRN Working Paper 3378076, co-authored with Samantha Washington [SSRN]
Using a panel survey which tracks racial prejudice from 2008 to 2018, this research shows that different measures of prejudice declined or at least did not increase in the first two years of Trump’s presidency. [Media: FiveThirtyEight, Washington Post, Philadelphia Inquirer, Spectator]
Journal of Politics, 2020, 82(2):529-542; SSRN Working Paper 2799347, co-authored with Seth Goldman
Extensive research suggests that non-Hispanic whites living near African Americans have higher levels of prejudice. In this paper, we show that where people lived as adolescents–not where they live as adults–predicts their levels of prejudice. [Appendix] [Data] [SSRN]
Journal of Politics, 2019 81(1):315-320 ; SSRN Working Paper 2798622, co-authored with John Sides and Jack Citrin
People who over-estimate the share of immigrants in the U.S. tend to be more opposed to immigration. In this paper, we use four experiments to show that providing accurate information about actual immigrant population shares does little to dampen opposition to immigration. [SSRN] [Appendix][Data] [Media: Washington Post]
Proceedings of the National Academy of Sciences, 2019,; co-authored with Seth J. Hill and Gregory Huber
This manuscript examines the relationship between local demographic changes and shifting voting patterns in U.S. presidential elections between 2012 and 2016. To do so, it employs novel, precinct-level data from six key states. [SSRN] [Data] [Media: Economist] [Award: Best Article, APSA Citizenship and Migration Section]
Journal of Experimental Political Science, 2019 7(2):112-136; SSRN Working Paper 3236348, co-authored with Cheryl Kaiser, Efrén Pérez, Sara Hagá, Corin Ramos, and Michael Zárate
Using five experiments with Asian American and Latino subjects, this paper reports no evidence of a causal link between perceptions of group-targeted discrimination and party-related attitudes. [Ungated, SSRN] [Data]
International Journal of Public Opinion Research, 2019 31(4):649-668; SSRN Working Paper 3268012, co-authored with Seth Goldman
Prior research has shown that exposure to Barack Obama in 2008 reduced anti-Black stereotypes among whites–but did he have the same effect in 2012, when we was a more familiar and partisan figure? This article shows even in the polarized environment of 2012, exposure to Obama reduced anti-Black prejudice. [Ungated]
Politics, Groups, and Identities, 2016, 4(3):363-92, co-authored with Jonathan Mummolo, Victoria Esses, Cheryl Kaiser, Helen Marrow, and Monica McDermott
Both scholarship and journalism lead us to expect that immigrants will experience different levels of discrimination depending on where within the U.S. they settle. To the contrary, this paper finds a striking lack of spatial variation in first-generation immigrants’ perceptions of discrimination, and discusses the implications of its results. [Ungated, SSRN] [Media: FiveThirtyEight]
American Journal of Political Science, 2015, 59(3):529-548, co-authored with Jens Hainmueller
This paper uses conjoint analysis, a tool from marketing, to test a range of hypotheses about Americans’ attitudes toward immigrants. Drawing on a two-wave Knowledge Networks survey, it demonstrates that Americans view educated immigrants in high-status jobs favorably, while they view those who lack plans to work, have previously entered without authorization, or do not speak English unfavorably. I summarized this research at The Washington Post’s Wonkblog here. [Award: 2013 Best Paper, Elections, Public Opinion, and Voting Behavior Section of APSA] [Ungated, SSRN]
British Journal of Political Science, 2015, 45(3):531-557; SSRN Working Paper 1879965
Do immigrants who speak Spanish or have darker skin tones provoke more support for restricting immigration? This paper uses two population-based survey experiments to demonstrate that they do not. Instead, an immigrant speaking with a pronounced accent induces more pro-immigration attitudes, likely because the accent is seen as a signal of his desire to assimilate. I summarized this research on The Monkey Cage here. [Online Appendix] [Data]
Annual Review of Political Science, 2014, 17:225-249, co-authored with Jens Hainmueller
This review summarizes and then discusses just under 100 studies of immigration attitudes undertaken over the past two decades. Consistently, immigration attitudes show little evidence of being strongly correlated with personal economic circumstances. Instead, immigration attitudes are shaped by sociotropic concerns about national-level impacts, whether those impacts are cultural or economic.
Political Communication, 2014, 31(3):421-455
The Spanish language is a flashpoint in discussions about immigration. Do brief encounters with Spanish shape attitudes toward immigration and integration? Coupling a survey experiment with voting data from California, this paper shows that Spanish induces different responses among Republicans and Democrats–and that its use on ballots can influence voting patterns as well. [Ungated, SSRN]
Politics, Groups, and Identities, 2014, 2(1):35-51, co-authored with Van C. Tran and Abigail Fisher Williamson
This paper explores whether brief exposures to Spanish can influence Americans’ immigration attitudes–and uses multiple experiments to show that the impact hinges on people’s prior exposure to Spanish. For those who hear Spanish frequently, even a subtle exposure to written Spanish induces anti-immigration attitudes. [Earlier Version] [Data]
Political Research Quarterly, 2012, 65(2):443-459
This paper uses the post-Katrina migration as a source of exogenous variation to explore the impact of changing demographics on a variety of political attitudes and behaviors. It shows that Baton Rouge and Houston respondents to the Katrina evacuees in different ways, with Baton Rouge residents growing increasingly opposed to public benefits and Houston residents becoming increasingly concerned about crime. [Data] [Award: Best Poster, 2007 Summer Meeting of the Society for Political Methodology]
American Journal of Political Science, 2011, 55(4):814-830
This manuscript uses two data sets to examine the impact of Spanish-language ballots. Exploiting a sharp discontinuity in the Voting Rights Act, it demonstrates that Spanish-language ballots increased turnout among Spanish speakers, and that they decreased opposition to bilingual education in California’s 1998 Proposition 227. This research was discussed by Voice of America News and New America Media. [Data]
British Journal of Political Science, 2011, 41(3):499-524
The political impact of people’s neighborhoods can change quite dramatically depending on what topics are salient in national politics. This paper uses panel data from the U.K. to demonstrate that living near immigrants shapes concern about immigration primarily when the issue is prominent in national debates. Survey data from the U.S. reinforce this claim. [Online Appendix]
American Politics Research, 2011, 39(2):344-379
Past research contends that ethnic and racial diversity dampens spending on a variety of public goods in U.S. cities. This paper takes a second look, using the most extensive data set on city spending from 1950 to 2002 available to date. It shows that the relationship between diversity and public goods is far weaker than previously suspected. In recent years, the only consistent influence has been to increase criminal justice spending. This post at Wonkblog draws on this research. [Supplemental Information] [Data]
American Political Science Review, 2010, 104(1):40-60
Testing a novel theory of neighborhood effects, this paper shows that immigrants are construed as threatening under two conditions: when immigrants suddenly arrive in a community and when national rhetoric or frames are available to politicize their arrival. This article was covered in Emerging Issues and the New York Times. [Data]
Journal of Politics, 2009, 71(1):160-177
Using a unique data set on tax votes in Massachusetts communities, this paper demonstrates that sudden increases in diversity can dampen the provision of public goods by reducing the number of long-term projects put before voters. The article was discussed in the Boston Globe and The New York Times as well. [Supplemental Information][Data]
Political Behavior and Public Opinion
American Political Science Review, Forthcoming, co-authored with Hans Noel
This paper uses four surveys of activists from 2016 and 2021 to measure perceptions about which senators and prominent politicians are perceived as especially liberal or conservative. [Data] [Ungated] [Media: FiveThirtyEight, Philadelphia Inquirer]
Journal of Politics, Forthcoming, 2022, co-authored with Susanne Schwarz and Anjali Chainani
We report the results of a field experiment in the 2019 municipal election cycle in Philadelphia in which iterative postcards from city officials increased general election turnout by 1.5 percentage points.
Forthcoming, British Journal of Political Science; SSRN Working Paper 3840201, 2021
This paper uses a population-based panel survey to investigate changes in public opinion 2007 to 2020. It finds alarming levels of affective polarization and racial prejudice, but little evidence of shifts in these attitudes that might have foreshadowed the 1/6/2021 threat to American democracy. [Ungated] [Appendix] [Data] [Media: Vox]
Science Advances, 2021, 7(17), co-authored with Seth J. Hill and Gregory Huber
Proceedings of the National Academy of Sciences, 2021, 118(4); co-authored with Marc Meredith, Anjali Chainani, Nathaniel Olin, and Tiffany Tse
Journal of Politics, 2021, 83(4):1800-1817; SSRN Working Paper 3366994, co-authored with Will Hobbs
Did the ACA’s exchanges or its market-based features generate policy feedback effects? In this paper, we employ a range of methods to uncover offsetting positive and negative policy feedbacks for different sub-populations. [SSRN] [FiveThirtyEight] [Ungated] [Data]
Public Opinion Quarterly; 2019, 83(1):123-34. SSRN Working Paper 20163769, co-authored with Kalind Parish
Using a difference-in-difference design and data on more than 60,000 Americans between 2010 and 2016, this paper shows that the Medicaid expansion increased approval for the Affordable Care Act after its implementation in certain states. [Appendix] [Media: [FiveThirtyEight, The Nation] [Pre-publication version]
Political Behavior, 2018, 40(3):681-709; SSRN Working Paper 20163769
Did frames like those about “death panels” influence public opinion on health care reform during the contentious 2009-2010 debates on the issue? This paper applies automated content analysis to press releases, television appearances, and open-ended survey questions to address that question. It illustrates the limits of framing effects, as the public’s core reasons for supporting and opposing the Affordable Care Act were evident in its language even before health care became a salient issue. I summarized elements of this research in the Washington Post and on The Monkey Cage. [Ungated] [Data]
Quarterly Journal of Political Science, 2017, 12(1):37-57; SSRN Working Paper 2863930; co-authored with Jonathan Mummolo
When citizens are exposed to an argument on one issue, does it influence their attitudes on others? This research uses a novel, population-based survey experiment to show that the answer is largely “no”–framing effects tend to be constrained to the issues discussed explicitly by the political rhetoric. [Ungated] [Appendix] [Data]
Research and Politics, 2017, 4(4):1-7; SSRN Working Paper 3025144; co-authored with Eunji Kim and Soojong Kim
Does the tone of newspaper coverage about the economy systematically lead or lag public perceptions of economic performance? This paper considers evidence from nearly half a million newspaper articles and television transcripts from 24 media outlets, along with 246,000 responses to the Survey of Consumer Attitudes. Especially for national outlets like the New York Times and Washington Post, the tone of coverage shifts after public opinion, an observation which places important limits on the extent to which these national outlets are actively shaping public opinion. [Data and Code] [Ungated] [Appendix] [FiveThirtyEight]
Journal of Empirical Legal Studies, 2017, 14(1):79-128; co-authored with Marc Meredith, Michael Morse, Sarah Smith, and Jesse Yoder.
This paper assesses the effect of a change in voter ID policies in Virginia implemented between 2013 and 2014. The evidence indicates that the strict photo identification requirements did not seem to deter substantial numbers of Virginia voters from casting ballots in 2014, possibly because of outreach efforts that accompanied its implementation. [Media: FiveThirtyEight, FiveThirtyEight Podcast] [Ungated] [Data]
Political Behavior, 2016, 38(3):713-746; SSRN Working Paper 2307631; co-authored with Michael Bailey and Todd Rogers
Does inter-personal persuasion work in a presidential campaign? In this paper, we evaluate a large-scale field experiment in which 56,000 Wisconsin voters were randomly assigned to persuasive contacts through canvassing, phone calls, and mail. The results indicate a backlash, as canvassed voters were less likely to complete a follow-up survey and less likely to support the candidate backed by the campaign. I summarized this research at The Washington Post’s Wonkblog here. [Open-Access] [Data]
Quarterly Journal of Political Science, 2014, 9(1):115-135, co-authored with Jonathan M. Ladd
Can the introduction of a new media outlet with a distinctive ideological perspective shape election outcomes? This paper uses individual-level data from the 2000 National Annenberg Election Study to show that the effects of Fox News access varied by party. For a summary on Wonkblog at The Washington Post, see here. [Appendix]
Journal of Politics, 2009, 71(3):769-781
Analysts have commonly discussed the “Wilder effect,” the gap between how black candidates poll and how they perform on election day. Testing these claims on all elections for Senator or Governor from 1989 to 2006, this paper finds that the Wilder effect disappeared for black candidates by the mid-1990s. This article was discussed in media outlets including ABC News, Newsweek, The New York Times, The Portland Press-Herald, The San Francisco Chronicle, The Washington Post, The New York Daily News, Time Magazine, and Science. [Data]
Public Opinion Quarterly, 2012, 76(1):50-70. SSRN Working Paper 1736742
Americans’ perceptions of economic performance are a powerful influence on Presidential approval and candidate choice, but past research rarely considers what influences these perceptions. It also has not considered how rising income inequality has shaped economic perceptions. In past work, snapshots from elections create the impression that these assessments of economic performance are influenced only by income growth among the wealthy. Examining more than 215,000 respondents over three decades, however, we learn that income growth among the poor is frequently more influential. [Data]
Social Science Quarterly, 2009, 90(3):744-764
Can contextual factors shape attitudes toward the poor? Synthesizing racial and political theories of contextual effects, this paper explores attitudes about who is to blame for poverty. It demonstrates that both an area’s racial composition and its partisan composition can influence respondents’ views about why people are poor. [Code]
Social Science Quarterly, 2009, 90(3):770-776
A reply, this article extends the original evidence that racial contexts–and not certain other contextual measures–shape attitudes toward poverty. It then presents new evidence from an exogenous demographic shock to reinforce the claim that racial contexts shape views of the poor.
Local and Urban Politics
Political Science Research and Methods, 2018, 6(4):697-714; SSRN Working Paper 2567294, co-authored with Lindsay M. Pettingill
At the presidential level, we know a lot about why voters support incumbents–but we know far less about retrospective voting in mayoral elections. Using a novel data set with 341 mayoral elections, the paper shows that incumbent mayors suffer at the ballot box when their city’s unemployment rate is high relative to the nation’s. I also summarized this research at FiveThirtyEight. [Data]
American Politics Research, 2012, 40(4):665-700, SSRN Working Paper 1632390, co-authored with Katherine T. McCabe
Mayoral elections between black and white candidates frequently generate charged rhetoric. But do black mayoralties pursue different fiscal or employment policies once elected? This paper shows that the answer is typically no, with the exception of police hiring and spending. [Supplemental Information]
Political Behavior, 2012, 34(1):79-101, co-authored with Thad Williamson
Suburban communities in the U.S. have sometimes been condemned for their lack of civic engagement. This article takes a critical look at such claims. It uses multilevel modeling and data on nearly 30,000 Americans to demonstrate that certain aspects of suburban design do dampen political participation–but that other aspects of suburban design have cross-cutting effects.
American Journal of Political Science, 2011, 55(2):326-339, co-authored with Elisabeth Gerber
Do Democratic mayors spend city money differently? This paper uses regression discontinuity design and a novel data set of big-city elections to show that they do–but only on areas under their direct control, such as policing and fire. [Supplemental Information] [Data]
Wal-Mart and other large retailers have been heavily criticized for their impact on American communities, but social scientists have been slow to explore just how real those impacts are. This working paper uses matching and hierarchical modeling approaches to estimate how large retailers reshape the civic life of the communities they enter.
Co-authored with Kirk Bansak, Jens Hainmueller, and Teppei Yamamoto
This paper provides a general framework for analyzing voter preferences in multi-attribute elections using conjoints. With this framework, we demonstrate that the Average Marginal Component Effect (AMCE) is well-defined in terms of individual preferences and represents a central quantity of interest to empirical scholars of elections: the effect of a change in an attribute on a candidate or party’s expected vote share.
Cambridge Handbook of Advances in Experimental Political Science, eds. James N. Druckman and Donald P. Green. Forthcoming, co-authored with Kirk Bansak, Jens Hainmueller, and Teppei Yamamoto
This chapter provides an overview of conjoint survey experiments and their use in political science. It also summarizes recent research employing conjoint designs and uses a new conjoint survey of preferences for 2020 Democratic challengers as a running example.
Political Science Research and Methods, 2021 9(1):53-71. Co-authored with Kirk Bansak, Jens Hainmueller, and Teppei Yamamoto
When administering conjoint experiments, how many attributes can researchers include? This paper shows that conjoint survey designs are surprisingly robust to the inclusion of large numbers of attributes. [Ungated]
Political Analysis, 2018 26(1):112-119. Co-authored with Kirk Bansak, Jens Hainmueller, and Teppei Yamamoto
This paper examines a potential breaking-point in conjoint survey experiments: how many tasks can respondents complete before fatigue leads to excessive satisficing? [Ungated]
ACL, 2017, co-authored with Daniel Preotiuc-Pietro, Ye Liu, and Lyle Ungar
This paper characterizes how Twitter users’ language varies depending on their political ideology. Only those on the political extremes tweet about politics with much frequency. [538 Coverage]
Political Analysis, 2014, 22(1):1-30, co-authored with Jens Hainmueller and Teppei Yamamoto
This paper demonstrates the potential value of conjoint analysis to political scientists, using examples about vote choice and immigrant admission to the United States. In doing so, it develops a set of statistical tools for drawing causal conclusions from stated preference data based on the potential outcomes framework of causal inference. [Data] [Award: Editor’s Choice designation, Political Analysis ; Miller Prize for Best 2014 Article in Political Analysis]
Legislative Studies Quarterly, 2013, 38(1):5-30; SSRN Working Paper 1643658, co-authored with Lee Drutman
This paper exploits the public release of more than 200,000 internal emails by the Enron Corporation to better understand the types of lobbying that Enron engaged in. Enron devoted only a small fraction of its attention to campaigns and elections, and was far more active in formally participating in bureaucratic processes than previous work might suggest. It also uses tools from automated content analysis to focus attention on political emails. [Award: Best Paper, Political Organizations and Parties Section of APSA, 2010]
Public Opinion Quarterly, 2010, 74(2):201-222, co-authored with Gary King
Anchoring vignettes are an increasingly common tool to reduce differences in how respondents interpret survey questions. This paper uses survey experiments to develop insights about how to design surveys with anchoring vignettes. It shows that placing vignettes prior to a self-assessment question can prime respondents, improving the measurement of the underlying concept. It also develops advice for survey questions that ask respondents to compare themselves to others. [Data]
American Journal of Political Science, 2010, 54(1):229-247, co-authored with Gary King
The explosion of blogs and other digitized text presents opportunities to measure public opinion in new ways. This paper presents methods of automated content analysis that give approximately unbiased estimates of the proportion of documents falling into pre-specified categories. It then applies these methods to data on thousands of web logs about the 2008 U.S. Presidential election. [Data] [R Package]
American Political Science Review, 2005, 99(4):623-631, co-authored with Beth Simmons
Owing to self-selection, estimating treaty effects is challenging, as countries are more likely to sign treaties if they expect to comply with them. This paper argues that treaties can both screen among states and then constrain those states that become signatories. It uses propensity score matching, and finds that signing onto the International Monetary Fund’s Article VIII has a marked impact on signatories’ behavior. [Data]
JAMA Health Forum 3(3):e220034, 2022, co-authored with David Yokum, Andrew Feher, Elana Safran, and Joshua Peck.
This paper presents results from a large-scale experiment in which 745,000 Americans who had considered health insurance plans on the ACA Marketplaces were sent behaviorally informed letters encouraging enrollment. Letters increased enrollment by 0.3 percentage points–or 7%.
American Journal of Health Promotion, 2022, co-authored with Aleksandra M. Golos, Syon Bhanot, and Alison M. Buttenheim.
This paper presents results from two survey experiments with samples of online Pennsylvania adults in May and October 2020. As opposed to partisan messaging, scientific messaging increased the odds of intending to vaccinate against COVID-19, an effect confined to Democrats in the October survey.
Journal of Behavioral Public Administration, 2020, 3(2), co-authored with Syon Bhanot
This paper presents descriptive statistics from two online surveys of approximately 2,000 Pennsylvania residents on social distancing and other policies to fight coronavirus. It demonstrates that there is no backlash to social distancing when that policy is identified as urged by elites generally or by public health experts. [Data] [Video] [Preprint]
Hurricane Katrina sent hundreds of thousands of evacuees to communities across the country. This report details how Arkansas used church-based networks to assist the evacuees, and compares that response to those in Houston and Baton Rouge.