Mass Media and Electoral Preferences during the 2016 US Presidential Race

Christopher Wlezien and Stuart Soroka

Research typically concentrates on how mass media influences voters’ perceptions and preferences.   This is not surprising – we know that election campaigns spend a good deal of time and effort trying to influence the public, and that most campaign information reaches voters through mass media.  But there also are good reasons for news coverage to reflect public opinion rather than affect it, some sociological and others economic.  This expectation may surprise some readers, but we do not regard it as original or especially contentious, just under-studied and so not well understood.  Our recent work explores this aspect of the relationship between media and citizens, focusing on the 2016 US presidential election.  Over roughly 200 days leading up to the November election, we examine the bi-directional relationship between vote intentions and news content.

Our analyses rely on two time series: (1) vote intentions and (2) media “tone.”  The first is based on the many trial-heat polls conducted during the 2016 election year, drawing on data archived by the Huffington Post.  Specifically, we rely on 308 separate national polls during the election year, 100 of which from after the unofficial Labor Day kickoff of the general election campaign.  We aggregate these data to create time series of macro electoral preferences.  The second time series captures the balance of sentiment in news coverage of the two main candidates.  We capture the tone of media coverage for each candidate from roughly 30,000 artices from nine major US newspapers during the election year.  The tone of coverage is then identified using the Lexicoder Sentiment Dictionary (LSD), a simple dictionary-based approach to sentiment analysis.  Using these data, we produce net Clinton-Trump tone in daily coverage.  The analysis relates these two measures, with a focus on whether and when media coverage leads and/or follows voter intentions.

We begin with a basic first-order autoregressive distributed lag (ADL) models relating the variables, where each variable is regressed on lagged values of both itself and the other variable.  This allows us to see whether one variable influences the other, independent of the latter’s own history.  Results show that media coverage does predict vote intentions, but also that those preferences predict the tone of coverage.  That is, the media appeared to both lead and follow the public during the 2016 campaign.

We then incorporate the effects of various campaign events into the analysis.  Doing so reveals that both media coverage and voter preferences reflected certain events, primarily the conventions and the initial Comey intrusion. Taking account of these events, media tone appears to have little impact on voters.  Indeed, the relationship between media and public opinion appears to have changed over the course of the election, and the final weeks of the campaign show no effect of media coverage on electoral preferences.  There similarly is no effect of electoral preferences on the tone of media coverage, at least over short time horizons.  There is some evidence of followership over longer time horizons (beyond several days), however. In sum, media coverage in 2016 appears to have followed electoral preferences at least as much as it led them.

Our analysis has focused on a single election in a single year.  What about presidential elections in other years?  What does the research tell us about the effects of election campaigns more generally?  Clearly, one cannot generalize based on an analysis of 2016 alone.  We nevertheless expect that similar analyses of news coverage will find instances in which media powerfully lead opinion and others where they do not.  Indeed, it may be that the relationship between media coverage and vote intentions varies over the course of campaigns, as seemingly was the case in the 2016 US presidential election.  It is for subsequent work to contemplate the broader role of news coverage – as both leader and follower – during election campaigns in the US and other countries.

Christopher Wlezien is Hogg Professor of Government at the University of Texas at Austin.  Further information about him and his work is available on his website.

Stuart Soroka is Michael Traugott Professor of Communication Studies and Political Science at the University of Michigan.  Further information is available on his website.



Does Partisan Self-interest Dictate Support for Election Reform? Experimental Evidence on the Willingness of Citizens to Alter the Costs of Voting for Electoral Gain

Daniel R. Biggers

That elected officials seek to change electoral rules and institutions for their own advantage is widely accepted by both academics and in public discourse. Such efforts are perhaps best illustrated by the decades-long partisan battle across states over the ease with which eligible citizens can exercise their right to vote. Debates over what identification one must show to vote, when and how one can vote, and the degree to which state governments should facilitate the voter registration process all divide officials along partisan lines based on how those changes might impact future electoral prospects.

Less clear is whether average citizens think about electoral policies in a similar manner. Previous research suggests they do: partisans often share the preferences for election laws of their party’s leaders, and those identifying with the party out of power are more favorable toward reforms that could improve their chances of future electoral and policy success. That work, however, fails to conclusively show that citizens engage in this type of evaluation, as their attitudes may instead derive from simple cue taking or a reflexive desire to alter the rules of the game under which they lost. Furthermore, a large literature suggests that citizens value procedural fairness, which should make them hesitant to support any reform that games the system too much for one side, even their own.

In my recent article in Political Behavior, I show that citizens can and do evaluate election rules based on how they impact their party’s future electoral prospects. To do so, I focus on three common reforms that directly influence how easy it is for citizens to register and vote: requiring photo identification to vote, permitting registrants to vote before Election Day (early voting), and allowing citizens to both register and vote on Election Day (Election Day registration, or EDR). Requiring photo ID to vote makes turnout more difficult by introducing an additional burden on political action (even if relatively modest for many registrants), while the other two reforms make voting more convenient by expanding opportunities to register and vote. I selected these policies because they are among those most frequently debated and implemented across states, as well as the fact that they tap into stated concerns about ensuring procedural fairness. As these reforms directly affect who can and will take part in the political process, to the extent that fairness concerns moderate a willingness to game the system in one’s favor, they should do so here.

I asked respondents in two survey experiments to report their level of support for each of these policies. In explaining the policies, I used either a generic description of the individuals affected by the reform (e.g., “some eligible voters”) or identified them as supporters of one of the two major parties. These latter conditions provided respondents with a clear understanding of the implications of adopting or rejecting the reform in question (in terms of how it would affect the participation of partisan allies or opponents). The results show that party members consistently consider their own partisan self-interest. Both Republicans and Democrats express greater support (opposition) for a reform framed as making voting easier (more difficult) for co-partisans than when that same reform is framed as doing so for their electoral opponents. In other words, citizen update their attitudes toward the rules governing how easily people can participate in the political process when informed about how those rules impact their party’s prospects at the polls.

These findings have several important implications. For one, public opinion is thought to restrict the extent to which officials can shape electoral institutions for partisan interest, but the results here suggest that this constraint may not be as strong as some claim. Despite professed concerns about institutional and procedural fairness, some citizens appear open to altering the ease of participation to improve their party’s prospects. To be fair, the changes in policy support are relatively modest in size and still often signal a high degree of support. That equity considerations do not completely mitigate the desire to obtain an electoral edge, however, raises questions about the limits to which partisans can be swayed to game the electoral system in their favor. More broadly, the results show that when provided with the necessary information, individuals choose to update their attitudes to better reflect their preferred electoral outcomes. As such, the future prospects of electoral reforms and their support among the mass public likely hinge at least in part on the extent to which they are viewed as advancing partisan interests.


Early Voting Changes and Voter Turnout: North Carolina in the 2016 General Election

Hannah L. Walker, Michael C. Herron, Daniel A. Smith


In 2013, the United States Supreme Court handed down Shelby County v. Holder. The landmark voting rights decision removed jurisdictions covered under Section 5 of the Voting Rights Act from preemptive federal review. Immediately after the decision, North Carolina passed the Voter Information and Verification Act. Among other restrictions, the legislation created strict voter ID requirements, eliminated same-day voter registration, and reduced early in person (EIP) voting.

Activists challenged the law, and in the summer of 2016 a federal district court struck it down for its racially disparate impact. In the wake of the court’s decision, though, North Carolina’s County Boards of Elections responded by making various changes to EIP voting. Some counties expanded, while others contracted, the number of days, hours, and locations voters could cast ballots prior to Election Day.

In our paper, “Early Voting Changes and Voter Turnout: North Carolina in the 2016 General Election,” we examine whether changes to EIP voting affected turnout in the Tar Heel State. More importantly, we are interested if the changes impacted race and party subgroups differentially. Leveraging the natural experiment of non-uniform changes across the state’s 100 counties, we assess the impact of the changes on turnout among Black and white voters.

Existing scholarship on the topic is mixed. Reducing barriers to voting is historically associated with increased turnout, especially among marginalized voters. Yet, measures that increase the convenience of voting might simply allow habitual voters to substitute traditional Election Day voting for a more convenient method. Some scholarship goes so far as to suggest that voting early may dampen turnout because it reduces the civic significance of Election Day mobilization efforts.

Empirical strategies to isolate the impact of EIP opportunities, however, often suffer from ecological inference problems when drawing on aggregate-level data, or social desirability and sampling biases when drawing on survey data. The use of administrative voter files improves upon the former strategies, but the specter of omitted variable bias haunts this approach.

To avoid these pitfalls, we leverage the uneven changes to EIP voting made across North Carolina’s counties. Drawing on administrative voting records, we compare turnout of registered voters residing in precincts on either side of a given county border. Rather than compare individual turnout in counties increasing EIP voting to counties decreasing or not changing EIP voting, we restrict our comparisons to border dwellers. Our logic is that voters living near one another (albeit in adjoining counties) are more comparable than they are to a randomly selected voter from elsewhere in the state. We assess the validity of this assumption using census data and find that it holds reasonably well.

Conditional on race and party, we then assess the impact of changes to EIP offerings with proportional differences in voting behavior from 2012 to 2016. We compare the turnout of Black democrats to other Black democrats, white democrats to white democrats, and so forth. We do this for changes to five aspects of EIP voting: overall hours offered, evening hours, Saturday hours, Sunday hours, and number of polling locations available. Our findings are mixed. In some instances, we find that the expansion of hours is statistically associated with increased turnout. In others, we find that EIP expansion is either negatively or not all associated with turnout. In short, we find that the impact is context specific. We find no evidence that the expansion of early in-person voting had no impact on turnout, much less that it demobilized voters.

The institutional variation of EIP voting across North Carolina’s 100 counties in the 2016 election offers an opportunity to evaluate the impact of changes made to this popular form of convenience voting. We are hesitant, though, to draw broad conclusions from our analysis. To the extent that turnout among racial and party subgroups changed in 2016, we find little evidence that changes to EIP voting opportunities were a decisive causal factor, neither expanding nor suppressing the vote. Why and how voters may overcome challenges to voting presented by changes to EIP offerings remains an important point of inquiry for future research. We hope our unique research design may help other scholars interested in trying to isolate the effects of institutional change on voting behavior.

The Effects of Partisan Trespassing Across Candidate Sex

Nichole Bauer

 Record-breaking numbers of women are running for political office in the 2018 mid-term elections. Many of these female candidates are Democratic women selling themselves as tough and aggressive fighters—traits conventionally associated with the Republican Party. Take for instance, Amy McGrath. McGrath is a Democratic woman running for a House seat in a conservative district in Kentucky. McGrath announced her candidacy with a video highlighting her military experience, especially her status as the first female Marine to fly in combat missions. McGrath’s campaign message may reflect a partisan trespassing strategy where candidates emphasize issues and traits associated with the opposing political party. The goal of such messages is to expand a candidate’s issue and trait competencies to include qualities of both political parties and to expand a candidate’s base of electoral support to include in-partisan and out-partisans. My article investigates whether female candidates can successfully engage in partisan trespassing strategies.

Trespassing on the issues and traits of the opposing political party is a relatively common campaign strategy among female and male candidates. I argue female candidates will have a more difficult time partisan trespassing for two reasons. First, voters associate both Democratic and Republican female candidates with issues and traits that fit into feminine stereotypes rather than partisan stereotypes. The influence of gender stereotypes means that voters may perceive a trespassing message as a gendered rather than a partisan strategy. Second, voters stereotype Democrats as the more feminine party and Republicans as the more masculine party. The intersection between gender stereotypes and partisan stereotypes can lead voters to evaluate the messages of female and male candidates of the same political party differently.

I conducted three experiments that tracked voter responses to female and male candidates who trespassed on partisan issues or partisan traits. Two of the three experiments presented participants with candidates with whom they shared partisanship. The third experiment varied whether participants saw a trespassing message about a candidate with whom they shared partisanship or who belonged to the opposing political party.

Partisan trespassing messages produce three outcomes. First, both female and male candidates can successfully expand issue and trait competencies to include opposing partisan qualities. Second, trespassing strategies undercut the partisan strengths of both female and male candidates. All candidates end up with a net issue or trait loss. In other words, voters associate trespassing candidates with fewer overall issue and trait competencies compared to when candidates use a partisan consistent message. Third, partisan trespassing messages attract out-partisan support for male candidates but not female candidates. Partisan trespassing messages disproportionately cost female candidates on favorability among both co-partisan and out-partisan voters.

These results show that trespassing strategies are not very helpful for female candidates. Trespassing female candidates will not only have a difficult time securing the support of co-partisan voters, but trespassing messages will not entice out-partisan voters. Voters perceive these messages through the lens of gender rather than partisan stereotypes. These results are particularly consequential in light of the 2018-mid-term elections which feature large numbers of Democratic women running in relatively conservative districts and states, such as Amy McGrath’s candidacy in Kentucky. Many of these candidates are trespassing on Republican issues and traits, but these strategies may not have the intended effect of mobilizing out-partisan support. It is not only Democratic women who lose with partisan trespassing strategies but Republican women do as well. The negative effect of partisan trespassing means that female candidates will have a more difficult time winning state and national offices that require attracting out-partisan support such as Senate races, gubernatorial elections, or the presidency.

The Differential Effects of Actual and Perceived Polarization

Adam M. Enders and Miles T. Armaly

Although there is disagreement over the extent of polarization among the American mass public, especially on matters of policy, a growing body of work shows that individuals perceive a wide gulf between their party and members of the out-party. For example, Americans assume that members of the out-party are farther from them on the liberal-conservative continuum than is actually the case. In other words, people perceive more polarization than actually exists.

A great deal of research examines the individual-level correlates of polarization, as well as the consequences of polarization for political attitudes and behaviors like voting, political participation, and trust in government. Yet, these studies often employ measures of actual polarization — the distance between one’s preferences from those of the out-party, as measured by policy preferences. Our work contributes to our understanding of the effects of polarization by considering how mere perceptions of polarization impact subsequent political attitudes and behaviors, and compares the effects of perceived polarization to actual polarization.

We begin our analysis by constructing individual-level measures of the two types of polarization using American National Election Studies data from 1972-2012. Importantly, these measures are grounded in the individual, which differs from many conceptualizations of polarization. Instead of determining how far apart an individual views the Democrats from the Republicans, we ask how far apart the individual views herself from the party with which she does not identify. We employ these measures to investigate three sets of relationships. First, we explore the individual-level correlates of each variant of polarization. Second, we examine the effects of perceived and actual polarization on orientations toward the government (i.e., trust and efficacy), voting, and participation in the political process. Finally, we ask how the two types of polarization differentially impact affective evaluations of the parties, candidates, and ideological groups.

Congruent with previous research, we find that perceived polarization has risen more sharply than actual polarization over time. We also find that certain political and demographic properties relate to the types of polarization in different ways. For instance, ideological strength is more strongly associated with perceived polarization than actual; elite polarization relates to perceived polarization, but not to actual polarization. Perceived and actual polarization also relate to substantive attitudes and behaviors in different ways. Perceived polarization is more strongly related to self-reported voting, participation, and trust in government than is actual polarization. We even observe countervailing effects of perceived and actual polarization on voting, such that perceived polarization is positively associated with voting, whereas actual polarization is negatively related to voting. Finally, perceived polarization is much more strongly related to animus toward out-party candidates, ideological groups, and the parties than is actual polarization.

Our results suggest that researchers should consider two things when conducting studies regarding polarization. First, whether the correlate or consequence of polarization theoretically relates to actual distances in policy preferences, or perceptions thereof. Second, whether measures of polarization anchored to the individual are more appropriate than differences in aggregate party preferences. Furthermore, our results have broad implications for an increasingly polarized world where politics is a largely emotional enterprise. While Americans don’t actually have great disagreements on matters of policy, such disagreements need not be present to pose a meaningful problem. Rather, the mere perception of political differences is sufficient to enflame disagreements and polarize. When individuals perceive great distances between themselves and the out-party, they feel more loathsome toward out-party members and their representatives, they self-report lower rates of voting, and trust the government less. This perceptual gulf increases with sophistication, strength of attachments to parties, education, and elite polarization — all characteristics that are either fixtures of the political world or normatively desirable. Moreover, perceptions, unlike actual differences in policy preferences, can be incorrect. That misperceptions seem to influence behaviors, orientations, and affect more than actual issues is striking.


The Good, the Bad and the Ugly: Do Attractive Politicians Get a ‘Break’ When They are Involved in Scandals?

Daniel Stockemer and Rodrigo Praino

Researchers often refer to the connection that exists in people’s minds between “beautiful” and “good” as an axiom. Good looking individuals generally obtain various advantages in life.  Just to name a few of these advantages, they more easily make friends, they receive pay premium at their job and they normally advance their careers faster.  In politics, this idea that “beautiful is good” is thousands of years old; the first reference we could find to this way of thinking dates back to Plato’s Lysis. It basically means that attractive politicians receive more votes than unattractive politicians at the ballot box. If “beautiful is good” then, mutatis mutandi, “ugly” must be “bad.” In fact, research does show that ugly politicians do get less votes at the ballot box.

Very similarly, “bad” politicians who get entangled in scandals while in office tend to lose electoral support, at least at the election immediately after the scandal becomes public. This makes perfect sense. If being “good” triggers a reward by voters, than being “bad” must trigger a punishment. The problem is that both attractive and unattractive politicians can act “bad” and end-up involved in a scandal. As a consequence, what happens when the “beautiful” and the “bad” find their expression in the same politician?

In our article, we show that attractive politicians do get a ‘break’ when they are involved in scandals. Using University students as naïve coders in a version of the truth-of-consensus method, we collected original data on the physical attractiveness of every member of the U.S. House of Representatives who has been involved in a public scandal between 1972 and 2012. We used these data to test the moderating influence of physical attractiveness in the relationship between scandal involvement and incumbents’ electoral success. In general, we find that attractive politicians can survive a scandal with very little or no electoral consequences. In the meantime, unattractive politicians tend to be punished more harshly for their transgressions. In more detail, our results indicate that incumbents deemed physically attractive by our coders who choose to run for re-election after a scandal breaks not only tend to get re-elected in large numbers but also tend to suffer very little in terms of total votes received. In other words, it seems that their physical attractiveness somehow protects them from the electoral consequences of being involved in a scandal.

But not all scandals are created equal, and voters do seem to differentiate between different types of scandals. For example, according to our analysis, candidate attractiveness plays the largest positive role in presence of sex scandals. In essence, attractive politicians can easily survive a sex scandal, while unattractive politicians tend to pay a very steep price in terms of votes for their sexual transgressions. In fact, based on our analysis, it seems almost impossible for a very unattractive candidate to recover from a sex scandal. We also find that voters are not willing to forgive financial transgressions, regardless of how a politician looks. In other words, bribery, corruption, misappropriation and misuse of funds are off-limits, even for the most attractive politician. This finding implies there are limits to the ‘breaks’ that attractive individuals receive. In particular, voters, seem to draw a line in the sand when it comes to scandals of financial nature.



Religion and the Extension of Trust

By Paul A. Djupe and Benjamin O. Hsiung

[This is a forthcoming paper at Political Behavior – the full paper can be viewed here.]

Religion has been at the forefront and at the back of social movements, at times castigated for its quiescence and at others lauded for its prophetic voice. Social scientists have long found that religion can provide institutional, attitudinal, and mobilizational bases for collective action, but of course it does not always fulfill those roles. Instead, religious effects are multivocal – they could prop up democratic politics directly, but in many cases do not. Why?

A foundational way to index the democratic role of religion is in terms of the trust that it fosters. Trust in others and trust in governmental processes are arguably necessary conditions for collective action to take place.

So, how does religion influence trust? We believe it would be a mistake to look for an unconditional link of religion to trust; indeed results over time have found a variety of links from some of the same religious tradition measures. Instead, we argue that religion effects on social and political trusts vary depending on the values communicated. These sets of religious values, which we aggregate into religious styles, are the central products in the religious economy. Congregations adopt religious styles in word and deed that range from inclusive (open doors, low boundaries to outsiders, few requirements of insiders) to exclusive (difficult access, high boundaries to nonbelievers, and requirements of believers).

The analogy to trust is clear as these value sets inform people about the potential costs of trust as well as whether making connections is a valued activity. Those congregations that lower boundaries and encourage bridge building also implicitly promote trust in others and institutions – they are engaged in extending the radius of trust. Those congregations that emphasize the presence of evil, the high wages of sin, and religious requirements to be a good person undercut the worth of working with others who are not well known – the radius of trust is short.

Using two datasets, we explore these notions in our forthcoming work in Political Behavior. The first, a cross-sectional dataset of 412 respondents from an online panel helps us assess whether elements of these inclusive and exclusive religious styles are linked to social and political trust judgments. Then we turned to Mechanical Turk to assess the degree to which these styles are flexibly implemented through a priming experiment. Participants were either shown a set of inclusive or exclusive religious style measures before responding to several social and political trust questions.

The results of the experiment, below, highlight that religious effects are dependent on continual communication processes. People need to be reminded of what their worldviews demand of them. Among religious respondents, priming inclusion activates a relationship between adopting religious inclusion and both social and political trust. When not primed, agreement with inclusion has no bearing on social and political trust among the religious. The expected opposite effect results from priming exclusion – religious respondents who adopt an exclusive style drop their social and political trust when primed. There are other distinguishable effects, but we detect nothing systematic.


Figure 1 – Interactive Effects of Experimental Activation of Inclusive and Exclusive Religious Styles on Political and Social Trust (Marginal Effects of Agreeing with the Treatments when Primed, 90% CIs)

Our intent was to highlight that there is little hardcoded in religion that would make democratic societies harder or easier to run. Instead, we should understand religion as a set of intentional communities with a range of religious styles that vary in response to people’s needs in context. People need to be reminded about the content and implications of their worldviews, which after all is the point of weekly sermons. There are other points of contact within congregations bearing on trust that are explored in the paper.

Some clear next steps involve thinking systematically about the social and political environment to understand how people from the same religious tradition can adopt quite different religious styles as a result. This variation should result in different levels of trust and hence collective engagement with society and government. Some of this work has been done in Europe (e.g., Traunmuller 2011), but the US is ripe for this sort of analysis.


Paul A. Djupe, Denison University Political Science, is an affiliated scholar with PRRI, the series editor of Religious Engagement in Democratic Politics (Temple), and co-creator of (see his list of posts here). Further information about his work can be found at his website and on Twitter.

Benjamin O. Hsiung is a 2016 graduate from Denison University where he studied Political Science and Communication. He now works in healthcare analytics at Epic Systems in Verona, Wisconsin.