Political commentator Walter Lippmann described politics in his 1922 book “Public Opinion” as “pictures in people’s heads,” perceptions born out of information people receive and interactions with others. Those pictures influence how people view each other, as well as members of other political parties. 

On Thursday, Jan. 23, Yanna Krupnikov ’02, a professor of Communication and Media at the University of Michigan, presented a talk titled, “How We See Each Other: Perceptions Across the Political Divide” in Rapaporte Treasure Hall. 

Prof. Greenlee (POL) introduced Krupnikov, who studied politics and graduated summa cum laude from Brandeis in 2002. A distinguished scholar specializing in topics of political news consumption and polarization, she has co-authored five books, including “Partisan Hostility and American Democracy,” “The Other Divide” and “Independent Politics.” The event was organized by the Politics department with the support of the Dean of the School of Arts and Sciences Jeffrey Shoulson. The Abraham Feinberg Educational Network for Active Civic Transformation was also an event co-sponsor. 

Krupnikov worked with James Druckman, Samara Klar, Matthew Levendusky and John Barry Ryan on a 2022 study examining people’s misperceptions of others’ level of political discussion and ideological intensity. They asked select participants how often they talk about politics and to characterize their ideology and had others estimate how often someone from another political party talks about politics and their ideological intensity. Krupnikov and her colleagues measured the level of political discussion on a three-point scale: rarely, occasionally and frequently. Of the group asked to self-evaluate how much they talked about politics, 50 percent of respondents said they occasionally talked about politics and 27 percent of them said they frequently discussed politics. On the other hand, the respondents who were asked how much others discussed politics thought that 30 percent of people selected “occasionally” and 64 percent chose “frequently.” Similarly, respondents overestimated how ideologically extreme  people were. Even though 38 percent of respondents described their political positions as extreme, participants guessed that 69 percent of respondents would fall under the extreme category. In reality, the majority of participants described themselves as moderate. 

According to Krupnikov, how people think about opposing political parties correlates with their level of animosity towards those other parties and affective polarization. As opposed to stemming from differing stances on policy issues, affective polarization means people dislike each other based on their political party. The relationship between affective polarization and perceptions is that negative feelings toward others with different partisan identities are based on how each side views the other; thus, how people perceive the other person influences how they feel about that person. 

Turning towards the most recent presidential election, Krupnikov discussed data she collected with Carlos Galina, a PhD student at the University of Michigan, and John Barry Ryan, an associate professor at the University of Michigan, before and after the 2024 election. The general goal was to observe people’s perceptions of other partisans’ beliefs and compare their answers to public opinion data to evaluate the accuracy of their perceptions.  

Respondents’ answers will be compared with data from the American National Election Studies, which shows what the public actually thinks. They planned their studies to occur at the same time that the ANES conducted their 2024 survey and informed participants of their methodology. Ultimately, the participants’ task is to estimate the ANES’ findings.

While Krupnikov’s studies covered a variety of topics, such as trust in the government, immigration and current events, she showed results for only one question. Respondents had to estimate what percentage of Americans, Democrats or Republicans (they were randomly assigned to one of these three groups) would offer each response option for the following question from the ANES survey: “How much do you feel it is justified for people to use violence to pursue their political goals in this country?” The options are “not at all,” “a little,” “a moderate amount,” “a lot” and “a great deal.” 

The ANES’ 2024 data was not available for comparison, so Krupnikov displayed ANES data from 2016 and 2020. She chose to discuss the question on justification of violence because the public’s responses for it in 2016 and 2020 remained very similar. For instance, in 2016, 81.97 percent of respondents stated that it is not at all justified to use violence to achieve political goals. Similarly in 2020, 82.9 percent said the same thing. Even though Krupnikov could not compare her results to the ANES’ findings from 2024 yet, the consistency of this question’s answers in past data makes it ideal to use for initial comparisons. 

All respondents, regardless of their political party identification, overestimated Americans’ acceptance of using violence for political goals, compared to Americans’ actual responses in 2016 and 2020. According to Krupnikov’s research, people are slightly less likely to overestimate when talking about their own party (i.e. Democratic respondents guessing how Democrats would answer). Respondents estimating the other party’s responses (i.e. Democratic respondents estimating Republicans’ responses) tended to perceive that the other party was more accepting of political violence than their own party. Republican respondents’ estimate for Republicans was the one closest to what the ANES’ surveys actually found. 

After explaining their findings, Krupnikov explored possible causes behind why people were overestimating the other party’s openness to violence for obtaining political ends. Polarization was a possible explanation, but in general, they calculated that people’s chance of overestimating did not increase as polarization increased. 

Krupnikov found more promising results when looking at political involvement as a source. She defined political involvement based on time spent on politics, how important politics is to the person and whether they feel frustration with others who don’t engage in politics the same way they do. The current media environment enables deep involvement in politics because it has the capacity to quickly and constantly provide news for consumption. Moreover, advancements in communications technology and online communities provide more forums for people to express their political opinions and connect with others. 

Examining political involvement is pertinent to perceptions in numerous ways. Firstly, higher involvement means that these people pay more attention to politics. More knowledge could mean more accurate perceptions. However, the type of information politically engaged people consume matters: if the news is partisan or emphasizes extreme cases, then people’s perceptions can become warped. Higher involved people also have a higher certainty that they know what’s best in politics. For example, Krupnikov’s research showed that when given a set of COVID-19 policies and asked how confident they were that their solution to a COVID-19 problem would be a good one, respondents who were deeply involved were more confident that their policy idea was effective. On the other hand, Krupnikov did not find this pattern with education; people who were more educated were actually less confident in addressing problems related to COVID-19. 

Krupnikov said that high levels of political involvement are strongly correlated with negative feelings towards the other side. Deep involvement signals that politics is important to a person and thus they are likely to feel frustrated when others do it wrong in their eyes. Additionally, highly involved people view more things as political. In another study, Krupnikov asked people to read a list of comments and determine which ones are political in nature. Some comments were political and others were drawn from a Reddit forum that banned political discussion. As a person’s political involvement increased, they were more likely to say a comment was about politics. Level of political involvement also influences what networks people are involved with and people they interact with, so those relationships can impact perceptions. 

As someone’s political involvement becomes more important to them, they start to think that people in the other party, as well as their own party, say that it is justifiable to engage in violence. People who were not at all involved in politics were the closest to predicting the ANES baseline. Krupnikov explained that misperceptions’ co-occurrence with involvement could be because deeply involved people are very certain they’re right, they have negative feelings towards the other side and they are impacted by their networks developed from political engagement. 

Krupnikov acknowledged that it would be impossible to causally test whether political involvement shaped perceptions. She would need to be able to change participants’ levels of political involvement  in a lab setting, which is not feasible. Instead, Krupnikov, Galina and Ryan use the term co-occurrence. However, they theorize that in a causal chain, involvement comes first, then animosity and then misperceptions. Moreover, there are thorny questions regarding perceptions, especially in establishing causality. A study could assume that someone dislikes the other side because they misperceive them, but it could go the opposite way. People can misperceive others because they dislike them. 

There has been research on whether fixing misperceptions can reduce polarization. However, Krupnikov questions whether perceptions should be fixed in the first place. She proposed that future research delve more into what perceptions actually represent. Could misperceptions represent honest mistakes from people who don’t understand statistics? Or do they represent people’s beliefs about what is emblematic of the other party’s behavior? Who has misperceptions can affect how easy it is to correct the misperception. For instance, people who don’t understand statistics and those who misperceive based on something that they care deeply about would require different approaches for changing their misperceptions.