An increasingly common methodological issue in the field of social science is high-dimensional and highly correlated datasets that are unamenable to the traditional deductive framework of study. Analysis of candidate choice in the 2020 Presidential Election is one area in which this issue presents itself: in order to test the many theories explaining the outcome of the election, it is necessary to use data such as the 2020 Cooperative Election Study Common Content, with hundreds of highly correlated features. We present the Fuzzy Forests algorithm, a variant of the popular Random Forests ensemble method, as an efficient way to reduce the feature space in such cases with minimal bias, while also maintaining predictive performance on par with common algorithms like Random Forests and logit. Using Fuzzy Forests, we isolate the top correlates of candidate choice and find that partisan polarization was the strongest factor driving the 2020 presidential election.
Moderates are often forgotten in modern research on the purportedly polarized and hyperpartisan American voters. Some have pointed out that many Americans appear moderate, but others contend that these apparent moderates are actually politically unsophisticated or conflicted extremists. We develop a method to distinguish between genuine moderates, inattentive respondents, and people whose policy positions are not well summarized by a single liberal-conservative dimension. We find that most of the respondents who give a mix of liberal and conservative answers to policy questions are genuine spatial centrists though almost 30 percent of survey respondents express policy views better described as inattentive or unconstrained than by a single ideological dimension. Having identified different types of survey respondents, we investigate their political behaviors. Moderates and those who don’t map onto the single dimension participate less in politics than liberals and conservatives. Even so, they are especially consequential for electoral selection and accountability because they are most likely to change their vote choices in response to candidate quality and ideology. Our results suggest a need for renewed attention to the middle of the American political spectrum.
The difference-in-differences (DID) design is widely used in observational studies to estimate the causal effect of a treatment when repeated observations over time are available. Yet, almost all existing methods assume linearity in the potential outcome (parallel trends assumption) and target the additive effect. In social science research, however, many outcomes of interest are measured on an ordinal scale. This makes the linearity assumption inappropriate because the difference between two ordinal potential outcomes is not well defined. In this paper, I propose a method to draw causal inferences for ordinal outcomes under the DID design. Unlike existing methods, the proposed method utilizes the latent variable framework to handle the non-numeric nature of the outcome, enabling identification and estimation of causal effects based on the assumption on the quantile of the latent continuous variable. The paper also proposes an equivalence-based test to assess the plausibility of the key identification assumption when additional pre-treatment periods are available. The proposed method is applied to a study estimating the causal effect of mass shootings on the public's support for gun control. I find little evidence for a uniform shift toward pro-gun control policies as found in the previous study, but find that the effect is concentrated on left-leaning respondents who experienced the shooting for the first time in more than a decade.
We test the relationship between historical immigration to the United Statesand political ideology today. We hypothesize that European immigrants broughtwith them their preferences for the welfare state, and that this had a long-lastingeffect on the political ideology of US born individuals. Our analysis proceedsin three steps. First, we document that the historical presence of Europeanimmigrants is associated with a more liberal political ideology and with strongerpreferences for redistribution among US born individuals today. Next, we showthat this correlation is not driven by the characteristics of the counties whereimmigrants settled or other specific, socioeconomic immigrants’ traits. Finally,we conjecture and provide evidence that immigrants brought with them theirpreferences for the welfare state from their countries of origin. Consistent with thehypothesis that immigration left its footprint on American ideology via culturaltransmission from immigrants to natives, we show that our results are strongerwhen inter-group contact between natives and immigrants, measured with eitherintermarriage or residential integration, was higher. Our findings also indicatethat immigrants influenced American political ideology during one of the largestepisodes of redistribution in US history — the New Deal – and that such effectspersisted after the initial shock.
The presence of a westward-moving frontier of settlement shaped early U.S. history. In 1893, the his-torian Frederick Jackson Turner famously argued that the American frontier fostered individualism.We investigate the Frontier Thesis and identify its long-run implications for culture and politics. Wetrack the frontier throughout the 1790–1890 period and construct a novel, county-level measure of to-tal frontier experience (TFE). Historically, frontier locations had distinctive demographics and greaterindividualism. Long after the closing of the frontier, counties with greater TFE exhibit more pervasiveindividualism and opposition to redistribution. This pattern cuts across known divides in the U.S.,including urban–rural and north–south. We provide evidence on the roots of frontier culture, identi-fying both selective migration and a causal effect of frontier exposure on individualism. Overall, ourfindings shed new light on the frontier’s persistent legacy of rugged individualism.
We study how officeholder gender affects issue accountability and examine whether constituents evaluate women and men legislators differently on the basis of their policy records. Data from 2008 through 2018 show that constituents’ approval ratings and vote choices in US House elections are more responsive to the policy records of women legislators than of men legislators. These patterns are concentrated among politically aware constituents, but we find no evidence that the results are driven disproportionately by either women or men constituents or by issues that are gendered in stereotypical ways. Additional analyses suggest that while constituents penalize women and men legislators at similar rates for policy incongruence, women legislators are rewarded more than men as they are increasingly aligned with their constituents. Our results show that accountability standards are applied differently across legislator gender and suggest a link between the quality of policy representation and the gender composition of American legislatures.
Is local attention a substitute for policy representation? Fenno (1978) famously described how legislators develop personal ties with their constituents through periodic visits to their districts and carefully crafted communications. Existing work suggests that such interactions insulate incumbents electorally, creating less need to represent constituents’ policy preferences. Surprisingly, this important argument has never been tested systematically. In this paper, I use data on senator travel and staffing behavior along with survey data from the 2011–2018 Cooperative Congressional Election Study to investigate this claim. In addition to showing that areas with important campaign donors are significantly more likely to receive resources, I find that local visits may decrease approval among ideologically opposed constituents. Furthermore, I find inconsistent evidence regarding the effectiveness of local staff. These results suggest that local attention does not always cultivate goodwill in the district. Under polarized politics, home style does not effectively substitute for policy representation.
Globalization and automation have contributed to deindustrialization and the loss of millions of manufacturing jobs, yielding important electoral implications across advanced democracies. Coupling insights from economic voting and social identity theory, we consider how different groups in society may construe manufacturing job losses in contrasting ways. We argue that deindustrialization threatens dominant group status, leading some white voters in affected localities to favor candidates they believe will address economic distress and defend racial hierarchy. Examining three US presidential elections, we find white voters were more likely to vote for Republican challengers where manufacturing layoffs were high, whereas Black voters in hard-hit localities were more likely to vote for Democrats. In survey data, white respondents, in contrast to people of color, associated local manufacturing job losses with obstacles to individual upward mobility and with broader American economic decline. Group-based identities help explain divergent political reactions to common economic shocks.
How do racially concentrated policy changes translate to political action? Using official election returns, the Cooperative Congressional Election Study, and original data on the unprecedented mass closure of schools in segregated, predominantly Black neighborhoods across Chicago, we demonstrate that those living in the communities affected (1) increase their attendance at political meetings; (2) mobilize in support of ballot measures to avert future closings; and (3) increase their participation in the subsequent local election, while decreasing their support for the political official responsible for the policy on the ballot—at a higher rate than every other group. These findings shed light on how groups that previously participated at the lowest rates go on to participate at the highest rates on community issues that matter to them. We develop a theory of place-based mobilization to explain the role of “the community” in acting as a site of coidentification and political action for marginalized groups.
The electoral connection incentivizes representatives to take positions that please most of their constituents. However, on votes for which we have data, lawmakers vote against majority opinion in their district on one out of every three high-profile roll calls in the U.S. House. This rate of “incongruent voting” is much higher for Republican lawmakers, but they do not appear to be punished for it at higher rates than Democrats on Election Day. Why? Research in political psychology shows that citizens hold both policy-specific and identity-based symbolic preferences, that these preferences are weakly correlated, and that incongruous symbolic identity and policy preferences are more common among Republican voters than Democrats. While previous work on representation has treated this fact as a nuisance, we argue that it reflects two real dimensions of political ideology that voters use to evaluate lawmakers. Using four years of CCES data, district-level measures of opinion, and the roll-call record, we find that both dimensions of ideology matter for how lawmakers cast roll calls, and that the operational-symbolic disconnect in public opinion leads to different kinds of representation for each party.
The premise that constituents hold representatives accountable for their legislative decisions undergirds political theories of democracy and legal theories of statutory interpretation. But studies of this at the individual level are rare, examine only a handful of issues, and arrive at mixed results. We provide an extensive assessment of issue accountability at the individual level. We trace the congressional rollcall votes on 44 bills across seven Congresses (2006--2018), and link them to constituent's perceptions of their representative's votes and their evaluation of their representative. Correlational, instrumental variables, and experimental approaches all show that constituents hold representatives accountable. A one-standard deviation increase in a constituent's perceived issue agreement with their representative can improve net approval by 35 percentage points. Congressional districts, however, are heterogeneous. Consequently, the effect of issue agreement on vote is much smaller at the district-level, resolving an apparent discrepancy between micro and macro studies.
We elaborate a general workflow of weighting-based survey inference, decomposing it into two main tasks. The first is the estimation of population targets from one or more sources of auxiliary information. The second is the construction of weights that calibrate the survey sample to the population targets. We emphasize that these tasks are predicated on models of the measurement, sampling, and nonresponse process whose assumptions cannot be fully tested. After describing this workflow in abstract terms, we then describe in detail how it can be applied to the analysis of historical and contemporary opinion polls. We also discuss extensions of the basic workflow, particularly inference for causal quantities and multilevel regression and poststratification.
Scholars and political observers point to declining labor unions, on the one hand, and rising white identity politics, on the other, as profound changes in American politics. However, there has been little attention given to the potential feedback between these forces. In this article, we investigate the role of union membership in shaping white racial attitudes. We draw upon research in history and American political development to generate a theory of interracial labor politics, in which union membership reduces racial resentment. Cross‐sectional analyses consistently show that white union members have lower racial resentment and greater support for policies that benefit African Americans. More importantly, our panel analysis suggests that gaining union membership between 2010 and 2016 reduced racial resentment among white workers. The findings highlight the important role of labor unions in mass politics and, more broadly, the importance of organizational membership for political attitudes and behavior.
How do citizens want to be represented by elected officials in an era of affective polarization? Contemporary narratives about American politics argue that people embrace elite expressions of negative partisanship, above and beyond representation on policy. Using three conjoint experiments, I examine how individuals weigh the relative value of substantive representation on issues, constituency service, and partisan affect. The findings challenge the notion that Americans are primarily motivated by their affective, partisan identities and demonstrate the value of policy congruence and service responsiveness in terms of perceptions of political representation. The implication is that people evaluate elected officials in ways that we would expect them to in a healthy, functioning representative democracy, rather than one characterized by partisan animus. Even if polarization is driven by “affect, not ideology,” citizens prioritize representational styles centered around the issues that matter to them.
Amid historically low response rates, survey researchers seek ways to reduce respondent burden while measuring desired concepts with precision. We propose to ask fewer questions of respondents and impute missing responses via probabilistic matrix factorization. A variance-minimizing active learning criterion chooses the most informative questions per respondent. In simulations of our matrix sampling procedure on real-world surveys as well as a Facebook survey experiment, we find active question selection achieves efficiency gains over baselines. The reduction in imputation error is heterogeneous across questions and depends on the latent concepts they capture. Modeling responses with the ordered logit likelihood improves imputations and yields an adaptive question order. We find for the Facebook survey that potential biases from order effects are likely to be small. With our method, survey researchers obtain principled suggestions of questions to retain and, if desired, can automate the design of shorter instruments.
Academics and political pundits alike attribute rising support for right-wing political options across advanced democracies to the working classes. In the United States, authors claim that the white working class offered unprecedented and crucial support for Donald Trump in the 2016 election. But what is the evidence for this claim? We examine all of the available academic survey data gathered around the election, along with a number of surveys from prior elections. We test four common claims about the white working class in 2016: (1) that most Trump voters were white working-class Americans; (2) that most white working-class voters supported Trump; (3) that unusually large numbers of white working-class voters switched from Obama in 2012 to Trump in 2016; and (4) that white working-class voters were pivotal to Trump’s victory in several swing states. We find that three of the four are not supported by the available data, and the other lacks crucial context that casts doubt on the idea that Trump uniquely appealed to working-class Americans. White working-class Americans have been supporting Republican presidential candidates at higher rates in recent elections, but that process long predates 2016, and narratives that center on Trump’s alleged appeal obscure this important long-term trend.
Theories of representation suggest that candidates should respond ideologically to their constituency. Two-stage elections like those in the U.S. force candidates to decide which parts of their constituency they should respond to: citizens who are active enough to participate in primaries or those who only participate in general elections. We posit that non-incumbent candidates should mostly focus on the preferences of primary voters while incumbents should be largely unmoved by the preferences of either set of voters. We test these expectations using data from U.S. House and Senate contests and find support for our theory. Our results suggest that scholars should pay closer attention to the two-stage nature of U.S. elections when evaluating electoral responsiveness.
Political commentators have offered evidence that the “polling misses” of 2016 were caused by a number of factors. This project focuses on one explanation: that likely-voter models—tools used by preelection pollsters to predict which survey respondents are most likely to make up the electorate and, thus, whose responses should be used to calculate election predictions—were flawed. While models employed by different pollsters vary widely, it is difficult to systematically study them because they are often considered part of pollsters’ methodological black box. In this study, we use Cooperative Congressional Election Study surveys since 2008 to build a probabilistic likely-voter model that takes into account not only the stated intentions of respondents to vote, but also other demographic variables that are consistently strong predictors of both turnout and overreporting. This model, which we term the Perry-Gallup and Demographics (PGaD) approach, shows that the bias and error created by likely-voter models can be reduced to a negligible amount. This likely-voter approach uses variables that pollsters already collect for weighting purposes and thus should be relatively easy to implement in future elections.
Imagine you’ve scooted into a red booth in an unfussy local diner somewhere in Michigan, not unlike those portrayed in the numerous articles reporters have dispatched from the Midwest since the 2016 election. One booth over, you’re overhearing a middle-aged white man talk about his politics with a buddy of his.
You find out over the course of your meal that he’s a moderate Democrat who wants to keep Obamacare protections in place and opposes concealed-carry, but who also supports mandatory minimum sentencing and favors deporting illegal immigrants. He also happens to mention that he voted for Donald Trump.
This sort of conflicted, “cross-pressured” voter often appears in vigorous debates over swing voters in quasi-hypothetical terms. However, we know from the Cooperative Congressional Election Study (C.C.E.S.), a nationwide scientific survey, that this aforementioned voter in Michigan is a very real, living, breathing man, who was among the roughly 65,000 Americans asked about their identities, policy preferences and voting behavior by the study.
Questions about people’s perceptions of politicians or other political actors are of central interest in a wide variety of research areas. But measuring these perceptions is difficult in part because respondents may use survey response scales in different ways. In a classic article, Aldrich and McKelvey (1977) introduce a method adjusting for such differential item functioning by assuming that all respondents perceive political stimuli identically. I propose a modeling approach built on the Aldrich and McKelvey framework but incorporating anchoring vignettes. This approach allows for scale use adjustments without assuming that all respondents perceive a given politician identically. I apply this model to data on Americans’ perceptions of parties, elected officials, and other political actors, showing that, contrary to previous arguments, most variation in ideology ratings is due not to differing scale use, but to differences in underlying perceptions. Specifically, while perceptions of Republican politicians and the Republican party show no significant differences by respondent partisanship, Democratic and Republican respondents differ strongly in their perceptions of the ideology of Democratic political actors as well as the Supreme Court.