Perceived Reciprocity and Democratic Threats

Paul A. Djupe, Denison University

Jacob R. Neiheisel, SUNY-Buffalo

Once reserved for groups on the political margins, questions of whether to extend basic rights and liberties have entered the mainstream. The last several election cycles were filled with rhetoric raising the stakes of control over the federal government by threatening that the other side will move to strip the losing party of, effectively, their freedom to organize. In one August 2020 rant, then-President Trump asserted that a Democratic administration would “[t]ake away your guns, destroy your Second Amendment. No religion, no anything, hurt the Bible, hurt God” (BBC 2020). This strikes at the heart of democratic politics which hinges on rights reciprocity – respecting others’ ability to express their grievances by voice and vote. Without a belief in reciprocity, there is little incentive to cooperate or afford the government any legitimacy. 

While reciprocity has been discussed in academic literature, it has had no explicit survey measure until now. In one particularly poignant example, Petersen et al. (2011) suggest that the reason some widely disliked groups are tolerated more than others is because they would reciprocate respect for others’ rights. They have no direct measure of this aside from the fact that tolerance of groups is still different even after controlling statistically for many determinants of tolerance though have pre-tests showing that some groups are perceived as being less likely to follow the rules of the democratic game. In effect, the remaining differences are the measure of a group’s reputation for respecting the rights of others. In this post, we assess this claim using a new measure of reciprocity that is set to debut in a forthcoming article in Political Behavior

Our measure of reciprocity is a simple one. After answering a least-liked group tolerance battery, we asked, “What if the tables were turned? Do you think that [your least-liked group] would respect your right to hold rallies, teach, speak, and run for office?” In several surveys, we asked this of only the least-liked group respondents selected. But in one we expanded the list to ask about the perceived reciprocity of a number of groups. 

Shown below in Figure 1, we split the results by partisanship to confirm how deeply divided the US is in 2020. Partisans perceive the lowest levels of reciprocity from their least-liked group (“L”), but the other party is not too far ahead and both are under water – Republicans believe Democrats (“D”) are not likely to respect their rights (the score is under .5) and Democrats believe about the same about Republicans (“R”). Conversely, partisans expect the most reciprocity from members of their own party. And, as has been often remarked, there is an asymmetry in how culture war groups are viewed – Democrats expect more reciprocity from “conservative Christians” than Republicans expect from atheists. 

Figure 1: Responsiveness of Expected Reciprocity to Partisan Ingroup/Outgroup Status. Source: 2016 Panel, June 2017 Wave. Note: A=Atheist, C=Christian conservative, D=Democrats in Congress, L=Least-liked group, R=Republicans in Congress.

Knowing the extent to which voters’ concerns for their rights have been activated may be consequential to their political behavior. But we are also care about how to classify and catalog the impact of this measure – is it distinct from existing concepts of group prejudice and threat?

We pursued these questions in several ways. It is not surprising that perceived reciprocity is correlated with both prejudice toward and threat expected from a group. It would be strange if it wasn’t. But reciprocity is much more strongly correlated with an existing threat measure – we used an index asking if the group was a threat to the respondent personally, to other people’s freedoms, and to the American way of life. Reciprocity is more strongly correlated to the latter two measures than it is to personal threat. In a model of political tolerance, we find that expected rights reciprocity is an important determinant with the same impact as variations in prejudice or education. It has a much stronger effect when the threat index is not in the model, but still has a statistically significant effect when the threat index is included. We take this evidence to mean that reciprocity is a heretofore unmeasured component of threat.

Given that these results were from cross-sectional data, we sought some additional tests to gain greater confidence that the link between reciprocity and tolerance is not spurious or just a reflection of tolerance judgments. Our data from 2016 were part of a panel that included tolerance and reciprocity judgments in multiple waves. We found that people who switched least-liked groups between panels shifted reciprocity to match the group’s reputation. That is, reciprocity dropped when the group had a worse reputation for reciprocity. Moreover, the change in reciprocity from wave 1 in September to wave 2 in November produced a corresponding shift in political tolerance even after controlling for shifts in the threat index and in prejudice.

Importantly, measures of which group an individual chose as his or her “least liked” continue to do a fair amount of work in predicting political tolerance even after taking into account whether members of such a group would respect the civil and political rights of others. To us this suggests that tolerance judgments are motivated, at least in part, by group animus as well as by concerns over whether members of such a group would respect others’ rights. 

Taken together, these findings complicate simple versions of the narrative that a principled intolerance—intolerance of those who refuse to respect the rights of others—is at the core of the public’s unwillingness to tolerate members of certain groups. They also raise a host of questions surrounding how it is that groups gain (or lose) a reputation for not engaging in rights reciprocity. With mediated representations of most groups being the only interaction that the average member of the general public will have with a member of their “least-liked” social group, are reciprocity judgments subject to distortions? What would it take for a group’s reputation to shift?   

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

Jacob R. Neiheisel, is an Associate Professor of Political Science at the University at Buffalo, SUNY. He very rarely tweets anything of note, but his account can be found here just the same.

Facts Shape Feelings: Information, Emotions, and the Political Consequences of Violence

Aidan Milliff, Massachusetts Institute of Technology

What makes violence political? Accumulated knowledge from political science, psychology, and criminology suggests that violence frequently leads to outcomes like political mobilization, revenge-taking, or spirals of conflict escalation. Violence generates grievances and anger, which, depending on the study, are causing such diverse outcomes as cycles of retribution and revenge (Costalli and Ruggeri, 2015; Balcells, 2017, and others), increased participation in politics (Bateson, 2012), withdrawal and skepticism toward government (Webster, 2017).

How can we reconcile these mixed findings and develop a clearer understanding of the relationship between victimization, anger, and the political consequences of violence? In “Facts Shape Feelings: Information, Emotions, and the Political Consequences of Violence,” a new article in Political Behavior, I focus on the middle link in this chain. I show that there is underappreciated variation in the emotional experiences of people exposed to potentially traumatic violence: people’s experiences are dominated by different emotions (not always anger), which they focus on different targets. This variation influences whether people attribute blame for the violence to perpetrators as individuals, or to the state and other political actors.

In the article, I analyze 31 original interviews with family members of homicide victims in Chicago, IL and find 1) that survivors of similar trauma express different levels of anger at different personal and political targets, and 2) there is a pattern in whose negative emotions have more political consequences.

I show that knowing particular information about the circumstances surrounding a relative’s homicide is associated with increased expression of anger toward the perpetrator as an individual, and with decreased prioritization of anger and blame targeted at political targets like the police, the city government, or the county prosecutor. People who can assemble clear understandings (I use the term “cognitive clarity”) about the perpetrator’s identity, motive, and culpability, are far more likely to target their anger at the perpetrator of violence. People who lack information about one or more of those categories might still feel angry, but their anger is more likely to diffuse toward political actors like the police or city government, even if they know the police are not the ones who perpetrated the homicide of their family member.

In addition to identifying patterns in how/when victims of violence express political vs. personal grievances, the article makes two methodological contributions. First, by comparing widely-used survey-style measures of anger to open ended discussions about anger and blame, I show that previous studies may be 1) conflating anger in response to violence with anger as a personality trait, and 2) conflating anger experiences that are directed at different targets and have different consequences for the angry person’s attitudes and behavior. In order to understand the political consequences of anger, I argue, future studies should at least measure blame when they are measuring anger.

Second, I demonstrate that unsupervised text analysis tools like topic models can be effectively used to present high-level patterns in even small corpora of sensitive text like the transcripts of in-depth interviews. I fit structural topic models (STM) to transcripts of my interviews in order to show patterns in respondents emotional experiences, attributions of blame, and access to information about their relative’s homicide. I take advantage of STM’s reliance on “bag of words”-style pre-processing to share replication data that makes the paper’s analysis reproducible, without sharing underlying natural language transcripts which are potentially identifiable and contain sensitive information.

The importance of cognitive clarity and the existence of variation in the emotion experiences of victims might extend far beyond Chicago. Evidence from Chicago shows that compelling “structural” narratives may not always match with the experiences of individual survivors of violence. One compelling way to understand homicide violence in Chicago, for example, is as a consequence of racism, structural inequity, and the “race-class subjugation” of communities where most deadly violence takes place (Weaver et al., 2020). Macro-level political narratives can be largely valid (Richie, 2012), but not all survivors of violence will use them to make sense of their experience. Even in other contexts where social scientists know why violence is happening, we should expect survivors’ responses to be highly variable and dependent on particular set of information that they can or cannot assemble to explain what has happened to them.

Do Public Consultations Reduce Blame Attribution? The Impact of Consultation Characteristics, Gender, and Gender Attitudes

Anthony Kevins, Loughborough University

Barbara Vis, Utrecht University

Politicians regularly have to make decisions that may end up backfiring. As a result, evading blame for any negative consequences of those decisions is a key goal for elected representatives – especially if they’d like to be re-elected. 

One way to try to minimize potential blame is to bring other people – such as opposition politicians, experts, or voters – into the decision-making process. In the current context, we need only think of the prominent role given to public health experts by prime ministers and presidents to get a sense for how popular this approach might be. But notwithstanding any inherent benefits and drawbacks of collective decision-making, doing so may help politicians by making it harder for citizens to pin down any responsibility later on. 

Building on this logic, public consultations – gatherings organized to solicit constituent opinions – may be an effective tool for politicians looking to avoid blame. But do they actually affect blame attribution? And if they do, what sorts of factors shape how effective consultations are at pre-empting blame? We examine these questions in a new article in Political Behavior, which uses two pre-registered survey experiments on nationally representative US samples to investigate the effect of: (1) consultation characteristics, e.g., whether or not the politician aligns their decision with constituent opinion; and (2) politician and constituent characteristics, e.g., the politician’s gender and constituents’ gender attitudes. 

In Study 1, we presented respondents with a vignette describing a city council decision to cut funding for a homeless shelter, which ultimately resulted in the death of two homeless people during a cold snap. Respondents randomly read either about a decision taken unilaterally by a mayor (the control), or a post-consultation decision the mayor took that either went against constituent opinion (listen and explain) or was in line with it (listen, explain, and align). We also considered, among other factors, the effect of varying the mayor’s gender (male versus female). Next, we presented the experimental treatments and outlined the potentially blame-worthy consequences. Finally, we measured our outcome: blame attribution. 

Study 1 – Predicted Values of Blame Attribution, Overall Effects and by Mayor’s Gender   

Note: All figures show mean responses with 83.5 percent confidence intervals, such that a lack of overlapping confidence intervals indicates statistical significance at the p < 0.05 level. Coefficients and levels of statistical significance (indicating differences relative to the corresponding control group) are also included. * p<0.05; ** p<0.01; *** p<0.001

Findings suggest that consultations shape blame attribution, but that gender does not. Panel A shows that respondents’ blame attribution levels were lower when the mayor listened, explained, and aligned their position with consultation participants’ preferences. Yet when the mayor’s position did not align with the majority, there was no such effect. Moreover, Panel B shows that there is no effect of the mayor’s gender in and of itself – a result that is consistent regardless of respondents’ hostile sexism levels. 

Study 2 then replicated Study 1 with a few important changes: it focuses on an issue that is less partisan-colored (flood-damage prevention instead of homeless shelter funding); it included more gender cues, to ensure that respondents were noticing the mayor’s gender; and it incorporated a distinction between incidental alignment (i.e., participants at the public consultation happened to agree with the mayor’s plan) and active alignment (i.e., the mayor changed course to align their plan with participant opinion). Results broadly replicated those of Study 1, so we focus here on Study 2’s novel components. 

Study 2 – Predicted Values of Blame Attribution, by (Non-)Adaptation and Mayor’s Gender

Does it matter whether the mayor changed course or simply pushed forward with their initial proposal? Panel A suggests that it does, since respondents tended to assign less blame to politicians who changed course – yet this is the case regardless of whether or not they consulted the population. Our results also reveal substantial variation where the politician’s decision did not align with constituent opinion: within the listen and explain treatment group, active dis-alignment (i.e., changing course where the population was supportive of the mayor’s initial stance) was linked to greater blame relative to the control, whereas we find no effect in cases of passive dis-alignment (i.e., staying course even though the population disagrees with the mayor’s stance). 

Finally, it’s also worth noting that despite the stronger gender cues in Study 2, we still find no evidence that blame attribution increases when the mayor is a woman rather than a man. As Panel B illustrates, we see only modest differences in consultation effects based on the mayor’s gender. Further corroborating Study 1’s results, Study 2 also finds little evidence of a connection between hostile sexism and blame attribution. 

Overall, our two studies show that public consultations may well, under specific conditions, reduce blame attribution – thus helping politicians to potentially avoid being blamed for decisions that end up backfiring. As long as participant opinion isn’t actively opposed to the mayor’s final decision, people seem less inclined to blame an elected representative who holds a public consultation on the issue.

Interestingly, this suggests that although public consultations are regularly praised for their potential to improve democratic quality, they may also serve as an effective anticipatory blame avoidance strategy: by muddying clarity of responsibility, voters may be less inclined to point the finger at the politician who took the decision. It is also striking that these dynamics seem to play out similarly regardless of the elected representative’s gender and the respondents’ gender attitudes. Our findings thereby broadly align with research suggesting that gender-related characteristics may have less widespread effects on reactions to scandals and controversies than is sometimes assumed. 

You Must Pay No Matter What: Moral Taboo and International Debt

Alessandro Del Ponte, Yale University

Peter DeScioli, Stony Brook University

And forgive us our debts, as we forgive our debtors.

Matthew 6:12, King James Bible, 1611

In an economic crisis, must a government repay its international debts even if it means cutting programs when citizens need them most? Governments struggled with this dilemma during recent crises in Greece, Argentina, Iceland, and other countries. The question is not only about economics: costs, benefits, credit, reputation. It’s also about morality: the obligation to repay and the obligation to care for citizens. 

When emboldened by moral righteousness, people stray far from ordinary considerations like costs, benefits, harm, compromise, and happiness. That’s why in so many cultures moralists condemn, punish, and even execute people for benign offenses such as premarital sex, dancing, homosexuality, eating the wrong animal, believing in the wrong gods, drawing the wrong cartoon, and so on. Human morality seethes with taboos—actions that are forbidden or obligatory, no matter what. And taboos collide to create dilemmas: Should you lie to protect someone you love? Should you take money from the rich to help the poor? Should you kill one person to save five?

We wanted to know how people judge moral dilemmas of international debt. How strong is the moral taboo against defaulting on a debt? To find out, participants read a hypothetical dilemma where a country suffers from an economic crisis and the government cannot repay its international debt unless it cuts government jobs. If the government defaults, then the lender country will have to cut jobs due to the missing funds but not as many as the debtor country. The dilemma is analogous to the trolley problem from moral philosophy, in which someone must decide whether to kill one person to save five people. In this case, the debtor decides whether to default on the debt to save 6,000 jobs at home, while causing 5,000 job losses in the lender country. This means that the debtor stands to lose 1.2 times more jobs than the lender, creating a dilemma between saving more jobs versus meeting the moral obligation to repay. Participants answered what the government should do, and whether each choice is morally wrong and deserves punishment.

We found that 72% of participants opposed default and said the government must repay (Figure 1, 1.2x condition).  Thus, we see the signs of a taboo: Participants opposed default even when it would do more good.

To gauge the strength of the taboo, we varied the debtor’s job losses across conditions. The debtor would lose 6,000, 10,000, 25,000, or 100,000 jobs, while holding constant the creditor’s losses at 5,000 jobs. This varies the ratio of damage to the debtor relative to the creditor, such that the debtor’s losses are 1.2x, 2x, 5x, or 20x those of the creditor. We also manipulated whether the government was deciding whether to fully default or to partially default by repaying half of what they owe. 

Figure 1. Percentage of participants who opposed default. Judgments are shown by the ratio of potential damage to the debtor relative to the lender and by whether the required default was full or partial. The line at 50% indicates the majority tendency; the percentages statistically differed from 50% except for the 2x/full and 5x/partial conditions.

Most participants still opposed default when the debtor’s job losses were twice the lender. As the debtor’s job losses grew to 5 times and 20 times the lender’s, participants became less opposed. But even when the debtor had 100,000 jobs at stake (20x), 39% of participants said the government must repay (Figure 1). Also, participants were less opposed to default when it was partial. This supports the idea that partially repaying shows the debtor’s good intentions and thus lessens the moral violation. More generally, this fits the idea that taboos depend on the nature of the action, including the intentions behind it, separate from the consequences.   

Participants’ moral judgments were different. Most participants judged that defaulting was wrong and punishable with little change as the debtor’s job losses increased. Many of these participants supported default even though they judged it to be morally wrong, presumably because they saw it as the lesser evil. In contrast, participants judged that the partial default is less morally wrong than the full default. These results show more signs of taboo: participants’ moral judgments depended little on the consequences but they did depend on the nature of the action, whether the default was full or partial.

In their comments, participants who opposed default emphasized the moral obligation to repay, using more deontic words: must, have to, need to, promise, commitment, contract, responsibility, agreement, obligation, fulfill, honor, owe, and duty. They wrote: “debts must be repaid,” “they have to honor the agreement,” ” and the debtor “should have to repay its debt no matter what.” Participants who supported default weighed the consequences of each choice, using more comparative words: less, more, better, worse, than, least, few, many, greater, number, and amount. Examples include: “I tried to minimize the job loss,” “I based it on the concept of greater good,” “Better for 5,000 people to lose their jobs than 100,000,” and  “I tried to get as few people affected as possible.”

We also look at political ideology by presenting the dilemmas to a national sample of Americans. Conservatives and Republicans were more stiffly opposed to default than liberals and Democrats. Still, most Americans across the political spectrum agreed that defaulting is morally wrong, even those who supported it. This suggests that liberals and conservatives share a moral code on debt, even if they adhere to it differently. 

Generally, these experiments illustrate how moral taboos can stand in the way of public welfare when they are at odds with better consequences. Taboos against default, amplified by ideology and partisanship, could thwart policies that would improve the greater good. Under the eye of morality, even economic issues like defaulting on a debt can become a sin to condemn at all costs.

Altruism and Spite in Politics: How the Mind Makes Welfare Tradeoffs about Political Parties

Alessandro Del Ponte, Yale University

Andrew Delton, Stony Brook University

Peter DeScioli, Stony Brook University

How much would you sacrifice to help your political party? When the news reports long lines at the polls, some voters persevere and some stay home. Similarly, when the party’s leaders falter—abandon principles, mishandle a crisis, commit crimes—some members endure the shame and carry on, while others turn against the leaders. Citizens also make tradeoffs to hurt the opposing party. When the opponents take our rights and flout our values, some citizens charge in protests while others cautiously let the anger pass. What’s the difference, and how do people make these tradeoffs? 

The answer, we suggest, is found in ratios, like the ones in music, painting, and mechanics. The ratios for parties are in the citizen’s mind: the ratio of the cost to the person compared to the benefit to the party. It’s the price you would pay to help or hurt a party, not only in dollars but also in time, effort, danger, and reputation. The mind uses this ratio to make tradeoffs, so it may be called a welfare tradeoff ratio, or tradeoff ratio, for short.

Tradeoff ratios are distinct from the mental processes found in traditional theories of partisanship, including theories rooted in attachment, identity, self-interest, attitudes, heuristics, and irrational emotions. For instance, theories about irrational attachments to a group do not say how these attachments factor into tradeoffs; the attachment would need to be a quantity stored in the mind if it is to be weighed against other quantities. The closest theories are about social preferences, such as preferences for fairness, altruism, and punishment. But tradeoff ratios are not general preferences that someone applies to everyone. We have specific ratios for each person and group, and they change as our relationships change.

The idea of tradeoff ratios comes from previous research on the psychology of interpersonal cooperation. The mind has special psychological abilities for making decisions that affect other people. Friends decide whether to help each other, parents decide how much to sacrifice for their children, and leaders decide whose interests to prioritize. To make these choices, the mind intuitively computes how much it is worth to help or hurt specific individuals, summarized by a cognitive variable, the tradeoff ratio. These computations are mostly unconscious, automatic, quick, and effortless, like the mind’s computations of color, depth, or the meaning of speech. People experience these computations as vague feelings, but the underlying cognition is precise and sophisticated, again like vision and language. To guide decisions, the mind uses tradeoff ratios to weigh another person’s welfare compared to our own.

So is helping political parties like helping people? To find out, we conducted experiments in which participants decided whether to sacrifice their own money to give or take money from a party. They made dozens of these decisions with varying costs in a random order. We wanted to know whether participants would make choices that are consistent with a distinct tradeoff ratio for each party, rather than choosing haphazardly based on vague feelings. In two experiments, participants’ tradeoffs were highly consistent with a distinct ratio (~90% consistent on average), despite making dozens of decisions in a random order. Moreover, participants’ ratios correlated in the expected directions with partisanship, political ideology, and feelings of enthusiasm and anger toward each party, corroborating that these ratios are politically meaningful.

In general, over 90% of partisans helped a party at a cost to themselves. And a substantial minority sacrificed money to spite their own party, illustrating the tension that some citizens feel toward their own party. Roughly 40% paid a cost (usually smaller amounts) to help the outparty, suggesting that these citizens feel at least some generosity toward both parties. At the other extreme, a quarter to a third of participants would pay any amount to harm the outparty, up to the maximum of sacrificing $100 to prevent the party from gaining $50 (more specifically, the chance of winning these amounts in a lottery). Overall, these findings echo other research finding that a fraction of polarized Americans coexists with a majority of moderate citizens. 

These findings may help transcend the traditional debates about whether voters are rational or irrational. Instrumental theories of partisanship emphasize narrow self-interest, but participants’ altruism and spite clearly contradict this idea. Expressive theories emphasize emotions and social identities, perhaps even suggesting that attachment to parties is irrational and uncalculating, which implies that people’s quantitative tradeoffs will be disordered and haphazard since they do not benefit from the precision of calculations. But in fact, participants’ tradeoffs remained highly consistent with a distinct cost-benefit ratio, even when challenged to make many tradeoffs in a random order. Participants’ altruism and spite toward parties were proportional to the costs and benefits, revealing that their tradeoffs did result from precise calculations, even if they didn’t maximize their own money. 

Thus, people’s partisanship may be rational in the sense of proportional to costs and benefits, rather than in the narrow sense of maximizing money. This idea echoes growing research in psychology finding that many emotions, including anger, compassion, fear, and disgust are regulated by precise (unconscious) calculations, which proportion our emotions to costs and benefits, such as more anger for more harm, more compassion for more need, and more disgust for more risk of pathogens. Underneath the turbulent surface of partisan passions hide precise calculations that proportion our altruism and spite toward parties.

Fascinated by ratios of all kinds, Leonardo da Vinci wrote in his notebook, “Proportion is not only found in numbers and measurements but also in sounds, weights, times, spaces, and in whatsoever power there may be.” Partisanship may be among the powers of society governed by proportions in the citizen’s mind. 

Preparation versus Relief: Understanding Public Sentiment Towards Natural Disaster Spending (Bechtel and Mannino)

blog post authored by Gianni Galasso, Michael Allison, Christine Hutchinson, and Holly Lawrence

This year, similar to last, the country will be hit by an above-average number of extreme weather events, including several major hurricanes. These natural disasters are known to cause immense economic and environmental damage. While the National Oceanic and Atmospheric Administration urges communities to act on these forecasts and prepare, the US government has for decades under-invested in disaster preparedness. According to a new study by Michael Bechtel and Massimo Mannino (forthcoming in Political Behavior), from 1985 to 2010 federal authorities allocated about 3% of all disaster-related spending to improving disaster preparedness, while 97% went towards disaster relief. This stark imbalance is extremely costly; existing estimates suggest that one dollar invested in preparedness is worth about $15 in mitigated future damage. The economic fallout caused by Hurricane Katrina, for example, could have been brought down from $100 billion to about $7 billion if the region had invested more aggressively in disaster preparedness instead of disaster relief. With such high costs, what explains this tendency to under-prepare?

Fig. 1 | Satellite imagery of Hurricane Katrina, which made landfall August 29, 2005 as a Category 5 Hurricane and took over 1,800 lives. Source:

One explanation for a lack of public support for disaster preparedness is that investment only seems worthwhile to those personally affected by extreme weather events. Plausibly, experiencing a natural disaster firsthand allows individuals to better understand the consequences of exposure to a disaster as well as to recognize a greater potential for future exposure. This experiential learning argument suggests that previously exposed individuals are more willing to invest in preparedness versus relief measures. Bechtel and Mannino’s study surveyed over 2,500 Americans, cross-referencing self-reported disaster exposure with geographic information. To their surprise, disaster exposure failed to predict support for disaster preparedness spending. In fact, individuals with medium and high levels of disaster exposure were just as unwilling to back preparedness investment as respondents with low affectedness.

What explains the absence of a systematic relationship between disaster exposure and policy preferences? Could mere exposure to natural disasters fail to provide respondents with information about the benefits of disaster preparedness spending, and perhaps even reinforce the belief that relief aid is the most effective approach because it is the most prevalent? Bechtel and Mannino explored this question using an experiment in which respondents were given information regarding the severity of disaster damages in the recent past and were asked about how they would split a $100 million budget between preparedness spending and disaster relief. They randomly assigned respondents into three groups. The control group received only the baseline information while the second group received a so called “compensation prime,” a short additional text about the government’s capacity to engage in relief efforts to compensate for damages and losses from a natural disaster. In contrast, the third group received information about the potential for preparedness investment to strongly reduce the damage caused by natural disasters.

Fig. 2 | Change in Disaster Preparedness Spending Relative to Control Condition in Percent. Results are broken down by treatment condition (no prime, compensation prime, damage reduction prime). The dots represent the change in preparedness spending as a percentage of the baseline level of preparedness spending in the control group. The vertical lines indicate 95% cluster-robust confidence intervals.

Figure 2 above displays the treatment effects observed in the compensation and damage reduction primes relative to the control condition. Compared to the baseline preparedness expenditure of about $50 million in the control group, represented by the green dot on the x-axis, respondents in the compensation condition were not willing to systematically invest more resources in disaster preparedness, represented by the red dot and confidence interval. This could mean that most people are already aware of the government’s ability to compensate for natural disasters with relief options, given that the presentation of this information had no effect on their spending preferences. When informed about the effectiveness of disaster preparedness, however, Americans changed their policy views and allocated 10% more funding to preparedness investment compared to the control group. This result, according to the research team, supports the idea that not knowing the economic benefits of preparedness over compensatory policy could help explain why personal disaster exposure and policy preferences are not systematically related.

The steep costs of under-preparedness, while the focus of Bechtel and Mannino’s study, are not unique to natural disasters. The ongoing COVID-19 pandemic continues to demonstrate the devastating effect of under-preparedness in the event of a public health crisis. While the pandemic came as a surprise for most people, infectious disease and other experts have been warning for years about an impending outbreak and our inability to handle it. In his popular 2015 TED Talk, former Microsoft CEO and founder Bill Gates noted that the US was seriously underinvesting and ultimately “not ready” for the next epidemic. While Gates’ concerns may not have garnered the public’s attention in 2015, they certainly have now. Over 600,000 deaths later, hindsight towards COVID –19 has initiated a strong push to prepare for pandemics in the future. Along with exposure to the pandemic came a flow of information about how better preparation could have saved countless lives and perhaps ended the COVID-19 outbreak in its infancy. As a result of learning about the effectiveness of preparedness measures, the public has displayed an increased desire for investing in pandemic preparedness in the future, supporting Bechtel and Mannino’s findings which show that when people are informed about the cost-saving measures of preparedness investment, policy preferences adjust accordingly. Educating the public about the efficacy of disaster preparedness may therefore be the most effective method in garnering its support and ultimately reducing disaster-related monetary and human losses in the long run.   

Rigid Religious Faith Promotes Selective Exposure to Attitude-Congruent Political Information

James Cragun, Stony Brook University

When people consume political information, they tend to selectively seek information that will support their prior attitudes and beliefs rather than information that will challenge them. For example, in the United States, Democrats prefer to watch MSNBC, while Republicans like to watch Fox. People who support gun-control policies like to read arguments that support gun control, while people who are against gun control like to read anti-gun-control arguments.

However, some people may do this more than others. Among two people who have equally strong opinions on a given issue, one person might choose to read things that are congruent with their attitude on that issue and avoid incongruent information, while the other person might seek information equally from both sides of the issue. Why is that? What could make one individual more likely than another to selectively seek attitude-congruent information?

There are probably many variables in a person’s life that could lead to stronger or weaker habits of seeking attitude-congruent information. One such variable might be religious faith. Some religions teach that maintaining one’s beliefs is virtuous and is even essential for gaining eternal rewards. Because of this, believers should be motivated to do things that will strengthen their existing religious beliefs. This could lead some religious believers to habits of selectively consuming information that will strengthen their beliefs and avoiding content that might challenge their beliefs. Though this habit may be developed in the context of religious beliefs, it could easily become a more general habit of seeking attitude-congruent information in other contexts, such as when seeking information on non-religious political issues.  

I tested this idea by having research subjects complete an information-search task on the topic of gun control. Subjects were given access to a list of pro-gun-control and anti-gun-control arguments, and I observed how many arguments they chose to read from each side of the issue. I found that subjects who had come into the study with more pro-gun-control attitudes chose to read more pro-gun-control arguments rather than anti-gun-control arguments, and subjects with more anti-gun-control attitudes read more anti-gun-control arguments rather than pro-gun-control arguments. However, among subjects who said they believe strongly in the importance of rigid religious faith, this pattern was even stronger. In other words, people who have firm religious convictions showed a stronger tendency to selectively read arguments that support their prior opinions even in the context of a non-religious political issue like gun control.

Of course, this doesn’t necessarily mean having rigid religious faith makes people more likely to seek attitude-congruent information. Rather, a tendency to seek attitude-congruent information could make people more likely to hold strong religious convictions. Or there could be some other variable that affects both these things.

To test whether religious faith directly affects political information-seeking behavior, I also included a randomized experimental treatment in the study. Immediately prior to the information-search task, the subjects completed a simple puzzle task in which they were instructed to unscramble sets of words to form sentences. For half of the participants, some of these sentences included words relating to religion (spirit, divine, faith, god, sacred, sermon). The other participants completed a similar task that did not include religious words. The purpose of this was to increase the salience of religion in the minds of half of the participants. This allowed me to see whether thinking of religious faith would produce different behaviors when seeking political information. It did. The tendency for pro-gun-control people to read pro-gun-control arguments (and for anti-gun-control people to read anti-gun-control arguments) was stronger among those who had been primed to think of religion. In other words, thoughts of religious faith increase the tendency to seek information that supports one’s prior opinions, even in the context of a non-religious political issue.

The results of this study do not mean that all religious people will have a greater tendency to read things that support their prior opinions. This study focused on one specific aspect of religiosity: rigid religious conviction. People who are frequently exposed to the message that religious convictions should be held firmly and not doubted (and who accept that message) are likely to exhibit a tendency to selectively seek attitude-congruent political information, but there may be some religious people who reject the importance of rigid faith and who are open to changes in their religious beliefs.  

Even when focusing specifically on rigid religious conviction, there were some individuals in the study who had rigid religious convictions but did not exhibit any tendency to seek attitude-congruent political information, and there were some non-religious individuals who had a strong preference for reading congruent information. As in most social-science research, the results should be interpreted only in terms of averages. Though religious faith does not all explain all (nor even most) of the variation between individuals, rigid religious faith tends to produce, on average, a stronger tendency to seek information that supports one’s prior attitudes.

The Gender Backlash in the Vote for Brexit

Jane Green, Nuffield College, Oxford University

Rosalind Shorrocks, University of Manchester

Increasing gender equality and economic competition between men and women represent some of the largest changes to society. These profound cultural changes are argued to be an aspect of the cultural backlash arguments for populism, alongside concerns about immigration and the economy.

We argue that men who perceive discrimination against men are more likely to support populist causes, even those where gender hasn’t been explicitly primed – such as in the case of ‘Brexit’. Our paper explores the predictors of male discrimination perceptions, assesses the economic grievance explanations for discrimination perceptions (finding that out of work younger males especially hold these perceptions in greater magnitude), and finds that these perceptions are associated with higher support for Britain’s decision to exit the European Union. Indeed, the effects of perceived male discrimination, among men, are as strong as other predictors based on discrimination about whites and ethnic minorities.

While the existing literature discusses the gendered nature of cultural change, the role of gender is under-explored empirically. It is argued that it is white working class *men* who feel left-behind and aggrieved about a loss of status, and yet the specific role of gendered discrimination perceptions has been so far largely ignored when trying to understand the political consequences of this ‘backlash’.

The clear exception to this has been the 2016 US presidential election and support for Donald Trump, where gender and gender discrimination attitudes played a very important role. However, that election was unique for many other reasons, not least the candidates themselves, and the priming of gender representation and divides.  

We therefore explore the question of gender discrimination perceptions in the case of Brexit. This is an important case, since gender wasn’t an explicit campaign focus in the British EU referendum. However, neither have voters’ motivations been argued to map onto the specific policy focus of the campaign. Rather the vote has been seen as a wider cultural backlash; a tide of resentment about social, cultural and economic change.

We devised new survey questions specifically to measure gender-discrimination perceptions. Crucially, these questions ask respondents to evaluate discrimination both in favour and against men, and separately, discrimination in favour and against women. Attitudes are not symmetrical, and neither do they show the same associations for men and women. Men are more likely than women to associate perceptions about male discrimination with perceptions about white Britons.

The existing literature would predict that lower income men would perceive greatest discrimination against men, or men with lower levels of education. We point instead to the importance of age and work status; whether men are in employment in sectors where they compete with women, and where these competitive experiences cause them to perceive gender-based discrimination. We find that it is especially younger men who are out of work who perceive greater discrimination, as well as to a lesser extent older pre-retirement age men. This chart shows the effects of gender, age and work status for men and women.

Not only do we find important differences in the relationship of these demographics with discrimination perceptions, we also find substantial and significant effects on Brexit support.

We expect that it should be men who are more likely to support Brexit if they perceive discrimination against men, since they are the group directly affected by this grievance. This is indeed the case. Furthermore, these gender discrimination effects among men are as substantial as the effects of perceptions of discrimination towards white British people and ethnic minorities – those groups who have featured much more prominently in existing explanations, to date.

The following graphs show the discrimination perception relationships on Brexit support, comparing these effects across types of perceived discrimination.

We should take gender far more seriously in theoretical and empirical work on the cultural backlash thesis, supporting other work that posits this effect (Norris & Inglehart, 2019). While the experiences of men have been highlighted by several influential works, we need a deeper understanding of why men demonstrate greater propensity to vote against the status quo, in support of populism, and on the basis of perceived discrimination against men. Changes with respect to gender equality have been as drastic as changes to the economy or racial diversity in established democratic societies, but the possible consequences of this have received much less attention. Our findings show that the gender backlash is widely relevant for political behaviour, even in a context where gender was not particularly salient, providing an important comparative example where this gender-backlash effect is in play. Overlooking gender discrimination effects overlooks an important phenomenon shaping recent political outcomes.

Partisanship, Religion, and Issue Polarization in the United States: A Reassessment

Jeremiah J. Castle, Metropolitan State University of Denver

Kyla K. Stepp, Central Michigan University

Does the public hold polarized attitudes on political issues? Despite the conventional wisdom that the public is deeply divided along partisan and religious lines, existing research has yielded conflicting results. In our article, we present a new theory of how social identities might influence issue polarization, as well as a new way of measuring issue polarization. Using an original survey fielded in February 2020, we find evidence that a sizable minority of our sample holds polarized attitudes on ten separate political issues. We find that partisanship predicts polarized attitudes on nearly all of these issues, but the impact of religious identities is strongest on cultural issues.

We develop a theory that the ability of social groups to impact political attitudes depends on individuals having the constraint to connect their social group identities to each specific issue. Applying this principle to partisan identities, we argue that partisan candidates, activists, and platforms have helped the parties stake out clearly defined, opposing positions on the issues. Given the easy availability of partisan cues across dozens of issues, we expect to find that partisan identities contribute to mass-level polarization on most issues.

Religious identities also bear the potential to lead to polarization, but religion’s effects ought to be strongest on cultural issues. In the second half of the 20th century, religious elites within the New Christian Right including Jerry Falwell Sr. and Pat Robertson took clear, unambiguous positions on cultural issues like abortion and same-sex marriage. As Castle shows in his book Rock of Ages (2019), during this period opposition to abortion became a key part of the evangelical subculture’s identity and distinctiveness. The clear messaging on cultural issues enables the public to connect their religious identities to attitudes on these issues. While religious elites have certainly taken positions on other sorts of issues, in the aggregate those positions have not been as clear or consistent, meaning that it’s less likely that religious identifiers in the mass public would have the tools to connect their identities to positions on these issues.

In order to test our theory, we wanted a measure of issue polarization capable of capturing the “depth” of polarization on each issue. Taking inspiration from the logic of the feeling thermometer difference used to measure affective polarization, as well as a question wording format previously used by the Pew Research Center, we fielded an original online survey in which we measured respondents’ attitudes on five cultural issues (abortion, same-sex marriage, teaching intelligent design in public schools, the Ten Commandments, and anti-transgender bathroom bills) and five non-cultural issues (social welfare programs, healthcare, immigration, the environment, and the size of the military). For each issue, respondents were asked to react to two statements (in a random order): one with a conservative frame and the other with a liberal frame. The questions were formatted like this (here we use abortion as an example, but the wording for all ten issues is included in our article):

How much, if at all, do you sympathize with those who say that in most cases a woman SHOULD be able to obtain an abortion, because she has a right to control her own body?

How much, if at all, do you sympathize with those who say that in most cases a woman SHOULD NOT be able to obtain an abortion, because abortion is murder?

Response options to each question were “A lot”, “Some”, “Not much”, and “Not at all”. By taking the respondent’s reaction to the conservative side of the issue minus their reaction to the liberal side, we generate a measure of issue polarization for each issue that ranges from -3 (as polarized as possible in favor of the liberal position on the issue) to +3 (as polarized as possible in favor of the conservative position on the issue).

We find evidence that a sizable minority of the population holds polarized attitudes on most of the issues presented. For every issue other than military size, more than 25% of respondents fall into one of the two polarized categories. The issue on which opinions were most polarized was same-sex marriage, where 46% of our weighted sample fell into one of the two most polarized categories.

FIGURE 1. Distribution of Issue Polarization Across 10 Issues. Source: Weighted 2020 Dynata data.

After controlling for a variety of demographics using regression, we find that partisanship is a significant predictor of polarization on 9 of the 10 issues we considered (the exception being teaching Intelligent Design, an issue that played a minor role in the “culture wars” but never really got much attention from the parties).

FIGURE 2. Party Identification’s Impact on Issue Polarization (with controls). Error bars represent 85% confidence intervals. Source: Weighted 2020 Dynata data.

The results for religion’s impact on issue polarization were mixed, closely matching our theoretical expectations. We find that religious tradition is a statistically significant predictor of polarization on cultural issues. However, on non-cultural issues the differences between evangelicals and the unaffiliated are typically substantively small and not statistically significant, once we control for other factors.

FIGURE 3. Religious Tradition’s Impact on Issue Polarization (with controls). Error bars represent 85% confidence intervals. Source: Weighted 2020 Dynata data.

Our findings contribute to the persistent debate over issue polarization by providing a theoretical framework for expecting when social identities might, and might not, contribute to mass-level issue polarization. In addition, our new empirical measure of polarization helps us capture the “depth” of issue polarization in a way that was not previously possible in the literature. We hope these contributions will be of use for researchers seeking to understand why politics in the United States has become more polarized, as well as what solutions might limit that polarization.

Intergenerational transmission of party affiliation within political families

Linuz Aggeborn, Uppsala University

Pär Nyman, Uppsala University

The accumulation of power within families has recently received increased attention from the academic community. A common objection against the Bushes, Cuomos, Le Pens, and Kennedys has been that they accumulate political power and consequently hinder candidates with new ideas. But is it true that children walk the same path as their parents?

In this paper, we investigate the intergenerational transmission of political affiliation using Swedish registry data. Our data cover everyone who has ran for local or national office between 1982 and 2014, but the analysis only uses those cases where both a child and its parent have been politicians. We have restricted the sample to candidates for the eight parties currently represented in the national parliament.

Note to figure: Illustration of a part of the main results, based on Table 1 in the published paper.

The probability that two randomly selected candidates belong to the same party is approximately 18 percent (it would be 12.5 percent if all parties were of equal size). It is therefore remarkable that around 76 percent of the children of a politician (n=14,275) choose to run for the same party as their parent. Given that Sweden is an egalitarian country with high income mobility, as well as a consensus-oriented democracy where the ideological distance between parties is relatively small, we would expect even larger effects in most other contexts.

To better understand this transmission we conduct a regression analysis where we control for the number of years that the child has lived together with the parent. This analysis shows that among children who never resided with the parent, only 31 percent run for the same party, highlighting socialization as the primary driver behind the strong intergenerational correlation.