Real Solutions for Fake News? Reducing Belief in False Stories on Social Media

Katherine Clayton, Spencer Blair, Jonathan A. Busam, Samuel Forstner, John Glance, Guy Green, Anna Kawata, Akhila Kovvuri, Jonathan Martin, Evan Morgan, Morgan Sandhu, Rachel Sang, Rachel Scholz-Bright, Austin T. Welch, Andrew G. Wolff, Amanda Zhou, and Brendan Nyhan

Social media has increasingly enabled “fake news” to circulate widely, most notably during the 2016 U.S. presidential campaign. Although it likely did not change the outcome of the election, false news threatens the democratic goal of a well-informed electorate. To combat the harmful effects of this misinformation, Facebook began adding “disputed” tags to articles in users’ news feeds that had been debunked by third-party fact-checking organizations in December 2016. (The company has since switched to providing fact-checks in a “related articles” format underneath suspect stories.) In April 2017 and May 2018, Facebook also promoted tips for spotting false news at the top of users’ feeds.

In our new article in Political Behavior, we investigate whether these and other types of misinformation interventions reduce belief in false stories on social media. Our preregistered online survey experiment tested the effects of three different types of warnings on belief in false news headlines. Participants recruited from Amazon Mechanical Turk were randomly assigned to read a series of true and false news headlines accompanied by “Disputed” tags (Facebook’s original approach), “Rated false” tags (a stronger warning of our own design), or no tags at all. Before doing so, some were also randomly assigned to read a general warning about misinformation on social media that was modeled after Facebook’s tips for spotting false news. Respondents rated the perceived accuracy of each headline as well as their willingness to “like” and “share” the stories on social media.

Our results show that both “Disputed” and “Rated false” tags reduce belief in false news headlines. The “Rated false” tag is more effective, but neither tag measurably reduces individuals’ willingness to “like” or “share” false stories. We also find that providing a general warning about false news modestly reduces belief in false headlines but also reduces belief in true headlines. Paradoxically, efforts to promote greater skepticism toward false news thus risk increasing public distrust of legitimate information.

The spread of “fake news” on social media remains an important concern today. Our findings indicate that specific warnings that alert readers about the credibility of the stories they encounter online are effective. Encouragingly, these results suggest that false news on social media can be countered with some degree of success if we are careful to avoid unintended spillover effects on belief in true stories.


Do Means of Program Delivery and Distributional Consequences Affect Policy Support? Experimental Evidence About the Sources of Citizens’ Policy Opinions

Vivekinan L. Ashok
Gregory A. Huber

Over the next two years we’re likely going to witness politicians propose a lot of new government spending. The gaveling in of the 116th Congress combined with the run-up to the 2020 presidential election means that voters are going to hear about policies that range from building green infrastructure to subsidizing college tuition. Research shows that how such policies are delivered—i.e., how the actual money is transferred to individuals and families—can affect support for them. Scholars of American politics have argued that politicians routinely obscure policies that benefit the wealthy by delivering them indirectly, through the tax code, in order to prevent them from being scrutinized by the public at large. But, what if citizens actually have their own reasons for preferring indirect policies?

In a study to be published in Political Behavior, we find that citizens have a strong preference that new government spending be delivered through the tax code and discover reasons for this point of view. Through a series of survey experiments, we ask citizens to consider two policy proposals that are intended to benefit most households: a program that provides all newborns and mothers with nutritional assistance for the first year after birth and a program that helps employed workers get additional training to remain competitive on the job market. While both programs would cover the costs of the respective benefits, we randomly vary the way in which each policy is delivered. In the “direct” version of the nutrition policy, households receive a debit card each month with funds to be spent on nutritional items; the direct version of the job training policy deposits money into workers’ checking accounts that must be spent on training programs or college classes. By contrast, the “indirect” version of both policies transfers funds as a tax credit. We find that citizens are substantially more supportive of both the nutrition and job training programs when the benefit is delivered indirectly, via the tax code. Respondents associate the indirect transfer of benefits with lower costs to the government, relative to the direct version of the policy. In regards to job training, citizens believe that the indirect policy will result in a smaller increase in their personal taxes compared to the direct transfer. And, respondents believe that the indirect version of the nutrition policy requires greater effort by beneficiaries than receiving a debit card each month.

Given that citizens favor indirect government spending when considering new policies, what are their views on current government programs? Does information about the distributional consequences—i.e., who benefits more—change the public’s support? To answer these questions, we examine support for two existing policies: the Home Mortgage Interest Deduction (HMID) and the Unemployment Insurance (UI) system. For each program, we randomly assign respondents to see information that frames the benefits as disproportionately favoring higher or lower income households. We find that for both programs, citizens favor expanding the benefits regardless of whether the policy is framed regressively or progressively. For the HMID there is no difference in support for the program between framings, whereas respondents show greater support for the UI system when its benefits are framed as favoring lower-income workers over higher-income workers. We do, however, find some evidence that the regressive framing of the HMID causes citizens to favor modifying the policy in a more progressive direction.

Our findings have practical implications for understanding how policy design can affect policy support. First, we show that citizens have a strong preference that new spending be passed through the tax code, believing it to be better than direct transfers for government’s bottom line. Second, we find no evidence that highlighting the regressive effects of current government policies will help garner public support for their removal—though proposals that make these policies more progressive, perhaps emphasizing marginal changes, could be popular.

Religion, Sectarianism, and Democracy: Theory and Evidence from Lebanon


Mohammed al-Amin Mosque, downtown Beirut

Michael Hoffman, Ph.D.

Does religious practice promote or dampen support for democracy? Critics of religion point to cases like Iran, where religious forces actively suppress pro-democracy movements, while supporters of religion cite examples such as Poland, where religious leaders and institutions helped undermine an authoritarian regime and pave the way for democracy. At the individual level, this issue is equally unsettled: does religion make citizens more or less supportive of democracy?

In a piece published recently in Political Behavior, I argue that religious attendance (or communal religious practice) can have both pro- and anti-democratic effects. I use the case of Lebanon, where religious-sectarian concerns strongly influence attitudes towards democracy, to illustrate how even within the same country, religious practice can have divergent effects on regime preferences. For Sunni Muslims, religious attendance has a robust negative effect on support for democracy. For Shi`a, however, exactly the opposite effect is present: those who regularly engage in communal prayer are considerably more supportive of democracy than those who rarely or never do so.

What explains these seemingly contradictory patterns? I argue that the missing link in this setting is sectarian interests. For Sunnis, who have benefited from the current political arrangement relative to Shi`a, the possibility of greater democracy is a threatening one: as a smaller and relatively more privileged group, they have a lot to lose from democratization. The larger Shi`a population, who have also historically tended to be neglected by the state, face precisely the opposite incentives. For them, democracy would mean greater representation, wider influence, and likely more access to state economic resources. Thus, the prospect of democracy looks very different to these groups, even within the same country.

Using original survey data from 2013-2014, I show that these relationships operate through a channel of group solidarity. The more individuals participate in communal prayer, the more they care about the interests of their group. For Sunnis, this translates into lower support for democracy, while it has the opposite effect among Shi`a. Respondents who frequently attend religious services—whether Sunni or Shi`a—consistently feel closer to members of their sect and see their political interests as tied to the interests of the group at large. In this sense, they view regime politics through a largely sectarian lens. Rather than focusing solely on private, individual concerns, they consider the implications that the prospect of democracy has for their group, and in doing so, adjust their regime preferences accordingly.


I use a variety of methods to test these claims, including experimental primes. When respondents were exposed to primes that activated “communal religious” concepts, they were more likely to report feeling a sense of sectarian solidarity. Likewise, they were more likely to report attitudes towards democracy consistent with the interests of their sect. Taken together, these findings suggest that communal prayer heightens the salience of religious group identity, and can noticeably affect attitudes towards key political issues in sectarian contexts.

Motivated Reasoning, Public Opinion, and Presidential Approval

Kathleen M. Donovan

Paul M. Kellstedt

Ellen M. Key

Matthew J. Lebo

The relationship between the economy and presidential approval is one of the largest and deepest literatures in political science. Scholars have debated between objective and subjective economic indicators, retrospective and prospective evaluations, and what factors moderate the relationship between the two, but their symbiosis has rarely been in doubt. The importance of the economy for assessments of the president are so fundamental that it even has its own catchphrase: “the economy, stupid.”

In a forthcoming article in Political Behavior, my colleagues and I contribute to a growing body of evidence that suggests this bond has weakened over time, another casualty of the rise in polarization. Citizens have always held multiple motivations when forming political opinions, including assessments of the president. Rising polarization, with its concomitant animosity and anger toward out-partisans, has lowered the importance of accuracy relative to partisanship when it comes to the motivations underlying presidential approval. The relative changing importance of these two motivations is abundantly clear given the spate of recent work investigating how to reduce partisan motivations when it comes to factual assessments.

Using time series data from 1981 to 2015, we find that presidential approval is increasingly untethered from economic assessments. Under the Reagan and Bush-14 administrations, in- and out-partisans rewarded the president in line with changes in the economy. Under Clinton, however, these effects became weaker, and disappeared altogether for out-partisans under Bush-43. Under Obama’s tenure, neither in- nor out-partisans responded to the economy; the two groups had made up their mind and the economy had not statistical impact on presidential approval for Democrats or Republicans. Preliminary signs indicate that the same is true of the Trump presidency, leading a number of pundits to cry, “It’s NOT the economy, stupid!”

On one hand, this finding might be celebrated: the president has long been held accountable for an economy that is composed of many moving parts, and of which the president controls only a few. If citizens were indeed assessing the president with a more nuanced understand of how the economy works and the role the president plays in it, our findings might be cause for celebration. However, as we argue, it is more likely that presidential approval has fallen prey to the partisan blinders that have affected so many of the public’s opinions. Moreover, these findings have implications for concepts closely associated with presidential approval, such as passing legislation, and presidential and congressional elections.

It’s Who’s on the Inside that Counts

Hans Hassell

Ever get the sense that the people who want to run for election are exactly the type of people you don’t want to have running for election? While ancient democracies occasionally used random selection to fill government positions, modern democracies fill those positions through a process of election of individuals who choose to stand for election. Those who choose to stand for election are significantly and substantively different from the general population. These differences are further exacerbated by the set of individuals who choose to work on political campaigns.

In “It’s Who’s on the Inside that Counts: Campaign Practitioner Personality and Campaign Electoral Integrity,” I examine whether the type of individuals who are interested in running for public office and who are participating in electoral campaigns are those who will foster a political environment that encourages widespread participation and a healthy democratic process. Specifically, this paper looks at how differences in individual personality traits of people working on campaigns affect the willingness of those individuals to use negative campaigning and to engage in unethical campaign behaviors.

Previous work has found that personality traits influence interest in running for political office and in engaging with political campaigns. This work takes the next step in looking at whether individuals with those traits that predict increased interest in political campaigns also incline those individuals to certain tactics on the campaign trail.

I find that they do. In this work, I use a survey of individuals working on senatorial, gubernatorial, and congressional campaigns to examine how personality affects campaign tactics and actions. I focus on three of the Big Five personality traits: agreeableness, extraversion, and conscientiousness. Individuals high in agreeableness have a higher concern for social harmony and trust other people. Individuals high in extraversion like to be noticed, like to take charge, and think they can influence others. Lastly, individuals high in conscientiousness plan carefully for the future and may be more likely to consider potential consequences of rash actions. In regards to political ambition, previous work has shown that those with higher levels of agreeableness and conscientiousness are less likely to express political ambition while individuals with higher levels of extraversion are more likely to be interested in running for public office.

In my work here, I find that that the traits that incline people to run for office are unfortunately also the traits that encourage campaign actions that may be deleterious to democratic institutions. I find that individuals with higher levels of agreeableness and conscientiousness are less likely to endorse campaign messaging that is negative in its tone. In addition, individuals with these traits are less likely to condone unethical campaign behaviors such as stealing yard signs or intentionally misrepresenting a candidate’s accomplishments. In contrast, individuals with higher levels of extraversion are more likely to engage in both activities. In short, the individual who are most likely to express interest in running for office (those with higher levels of extraversion and lower levels of conscientiousness and agreeableness) are exactly the type of individuals who are most likely to engage in objectionable campaign behaviors.

The Social Desirability of Rallying ’Round the Flag

Robert Urbatsch

Figuring out what someone really believes about politics and policy can be tricky. People often say things calculated aimed less at being true than at pleasing or impressing—or, in this era of polarized political trolling, provoking. Or they say what they themselves want to hear or what they find amusing rather than what they believe. While public-opinion surveys are invaluable, then, social scientists are always seeking additional means of corroborating survey responses.

Alternative, more honestly revealing measures are often elusive. Children’s names, though, may fit the bill. People select children’s names carefully, but typically only expect the first given names to have wide use. Middle names are more easily hidden. Monikers relegated to that position thus allow parents self-expression while making any related embarrassment easier to hide. Of course, not every name-giver thinks in such terms: some parents boldly give names without regard to social reactions, and others prefer to avoid controversy in any name. But middle names probably face weaker social-desirability pressures than do first names.

Since political events influence name choices—Election and Inauguration Days feature bumper crops of babies sharing the president-elect’s names, for example—this contrast between first and middle names can illuminate political attitudes that people hesitate to admit publicly. My new paper for Political Behavior examines how this plays out in moments when foreign policy becomes a hot political issue.

These moments of foreign-policy crisis typically create a “rally ’round the flag” effect, with surges in support for the government and its policy. But since going against the crowd at such moments carries an unpatriotic stigma, those who do oppose the popular position may hide their views in pursuit of social acceptability. To the extent this happens, rallies will appear to be larger than they really are.

Does this happen? Consider the run-up to the 2003 Iraq invasion in the United States. France’s failure to back America’s invasion led to public backlash, most famously in Congress renaming French fries as “freedom fries.” As might be expected, the prevalence of first names of French origin fell during the crisis, at least in the three states for which data is available. (This parallels falls in sales of brands with French-sounding names.)

The invasion, however, was deeply controversial, and many Americans may have secretly sympathized with the French position. This implies that their use of French-inflected middle names may not have fallen so much. Indeed, it did not: middle names of French origin did not become less common.

This contrasts with name changes around the 9/11 attacks. When terrorism was to the fore, more names echoed those of then-prominent politicians (“George,” “Rudy”) or had patriotic/stirring overtones (“America,” “Liberty”). But this was true for both first and middle names, suggesting, as might be expected, that more of that rally was sincerely felt.

Although not every political situation would feed into child-naming, this result fits into other stories of names revealing parental views. Because saddling a child with an undesirable name is costly, the difference between first and middle names can be exploited to gauge sincerity of beliefs, at least among those who name children, in ways few surveys can reach.

The Resistance as Role Model: Disillusionment and Protest Among American Adolescents After 2016

David E. Campbell and Christina Wolbrecht

University of Notre Dame

From white suits evoking the suffrage movement to her “I’m with her” slogan, Hillary Clinton framed her 2016 presidential campaign as an opportunity for women to break the “highest, hardest glass ceiling.” Clinton’s Fight Song promised: “We about to show the world, yeah, Women are equal now.” Many speculated that as a trailblazer, Clinton would be a role model for political activism among women and girls—as had previous women candidates.

In the end, however, the 2016 presidential election looked more like a setback to the promise of gender equality than a triumph. Clinton’s opponent had less political experience than she did, openly disparaged women, and was heard on tape bragging about sexual assault. Clinton failed to become president despite receiving a majority of votes cast nationwide—not to mention the credible claims about Russian interference. The 2016 election raised serious questions about the equality, representativeness, and integrity of the American political system.

What about those young women and girls who many expected to be inspired by Clinton? Was their faith in democracy dimmed? Did the controversial and sexist election contest dampen their interest in political engagement instead of encouraging it? To find out, we use a nationally-representative survey of adolescents aged 15-18 and their parents. Both answered questions during the campaign (early Fall 2016) and again after the campaign (early Fall 2017) about their views of democracy and political engagement.


The 2016 campaign had a clear impact. We find that after the 2016 election, girls who identify as Democrats became more likely to say that the American political system is not responsive to people’s needs—that is, they became disillusioned with politics. At the same time, all Democratic girls, but especially those who were disillusioned, became more likely to say that they will engage in political protest.


Why did Democratic girls—but not Democratic boys or Republican boys or girls—become interested in protesting post-2016? Rather than (only) being inspired by a woman politician, we suspect a different kind of role model effect is at work. A wave of activism—including the largest single-day demonstration in US history—emerged after the 2016 election. The majority of activists have been women and the protestors have used gendered rhetoric and symbols, such as the famous pink hats. Not surprisingly, adolescents (regardless of party or gender) who had a parent engaged in protest after 2016 were more likely to say they would protest themselves.


But this wasn’t only a family dynamic. Democratic girls have had other visible role models—women marchers and organizers in their communities and nationwide—for how to channel their political frustration after 2016, and many have come to see protest as an important part of their own political repertoire. Hillary Clinton was surely a role model in 2016, but the response to her loss by other women—in their own homes, communities, and country—has offered Democratic girls another model of how women can and should respond to political disappointment and frustration. If these effects persist, one lasting consequence of the Trump era may be a cohort of politically active women, whose entrée into politics can be attributed not only to inspiration but also to indignation.