Correcting Misperceptions of Fundamental Differences Between U.S. Republicans and Democrats: Some Hope-Inspiring Effects
General Discussion
Three experiments tested whether overlapping distributions that accurately depict similarities alongside differences between Republicans’ and Democrats’ values can correct misperceptions of differences and have positive downstream effects. We found consistently that participants perceived the groups to be more similar when they saw overlapping distributions, compared with seeing barplots or receiving no information.
Our other findings show why correcting misperceptions is important. Seeing truncated barplots performed similar to seeing no information, suggesting that the exaggerated depiction of group differences in truncated barplots is in line with people’s baseline perceptions. The social sciences and media may play a harmful role in fueling and perpetuating these biased perceptions through an emphasis on reporting group differences while ignoring the overlap. This emphasis is particularly problematic because it might set biased ingroup and outgroup norms that drive further polarization (Hartman et al., 2022). Group differences are often more newsworthy than commonalities, and although this focus can have its use and is unlikely to change, the overlapping distribution approach provides a method that can simultaneously display differences and similarities between groups for a more complete picture. Interestingly, this approach was also more effective in correcting misperceptions than untruncated barplots, which is a commonly recommended method for reporting differences in the social sciences (Yang et al., 2021). A reason for this greater effectiveness may be that distributions directly show the proportion of data points that are similar versus different, while additionally depicting the heterogeneity of both groups. Overlapping distributions may hence provide a more intuitive and comprehensive picture of group comparisons that translates into higher perceptions of similarity.
We found robust downstream effects of reduced misperceptions on qualitative expressions of hope; effects that were pre-registered in Experiment 3 and emerged consistently across three experiments and all 15 raters. Participants who received information on group overlap expressed more hope about the future of the country in open-ended comments compared with participants who saw truncated or untruncated barplots. This large effect was partially explained by a greater sense of shared reality between the groups and greater perceived potential for compromise. It was also fully mediated by qualitative ratings of how cohesive participants perceived U.S. society to be. Interestingly, these effects were robust across sociodemographic variables and order effects, suggesting widely applicable effects. Together, people appear to feel a sense of hopelessness because they see the country as deeply divided, and reducing the perceived division can help inspire new hope. These effects are particularly valuable in the face of growing pessimism in the United States (Najle & Jones, 2019; Pew Research Center, 2019a), and evidence that hope is a crucial driver for positive societal change and mental and physical health (Frumkin, 2022). Comparing groups that are commonly perceived as deeply polarized in overlapping distributions may hence be an optimistic signal that there is a way to find compromise and build a better future, thus eliciting greater expressions of hope.
It is surprising, however, that these robust effects on qualitative hope were not mirrored across two quantitative measures of hope. Experiment 2’s belief-based measure did not significantly correlate with the qualitative hope ratings, whereas Experiment 3’s affective measure showed a large correlation. There is clear overlap between qualitative comments that were consistently coded by raters as expressing high hope and the items used in both quantitative measures (e.g., “I feel very optimistic about the future”; see Supplement Table S13 for the comments). The findings suggest that although overlapping distributions robustly elicit unprompted expressions of hope, the effect is muddled when directly asking about hope, perhaps due to response biases or because participants consider other reasons for feeling hopeful or hopeless when prompted. Although the difference appears to be primarily methodological at first glance, with qualitative ratings showing the effect as opposed to quantitative measures, it should be noted that our measures of perceived potential for compromise may also be considered as reflecting a sense of hope, and these quantitative measures did show effects of the intervention. In any case, future research could take this evidence forward by testing whether expressions of hope also occur outside of written text, for example, in cross-party conversations, and whether conversations infused with hope are experienced as more pleasant, potentially eliciting more compromise and cooperative behavior.
The effects on perceived potential for compromise are further interesting because they extend to perceiving greater agreement on commonly divisive, concrete policies such as government spending on health and education. This finding may suggest that the intervention could open up conversations about progress on such divisive topics, rather than expecting deadlock and conflict that may thwart attempts from the start (Wald et al., 2024). Moreover, on this outcome, the overlapping distributions and untruncated barplots’ conditions showed comparable effects relative to truncated barplots. The reason may be that both methods provide a more accurate picture of the generally small group differences, and hence both methods could help improve people’s low expectations about cross-party interactions (Wald et al., 2024). Partisan animosity showed more mixed effects. There were no consistent effects on outgroup liking, but in Experiment 2, overlapping distributions elicited greater motivations to discuss with an outgroup member, and Experiment 3 found that overlapping distributions and untruncated barplots increased outgroup trust and reduced dehumanization. Although past evidence suggests that correcting misperceptions can reduce partisan animosity, these effects are generally small (e.g., Mernyk et al., 2022; Syropoulos & Leidner, 2023), and scattered effects are to be expected with small effect sizes (Lakens & Etz, 2017).
Recent evidence suggests that correcting some misperceptions can be more impactful than others, with information on outpartisans’ anti-democratic attitudes improving outcomes whereas information on outpartisans’ policy support or opportunism showed no effects (Voelkel, Stagnaro, et al., 2023b). As argued in the introduction, we would expect that discovering similarities in human values is particularly impactful because values are personally meaningful and relevant across a wide range of cultures and contexts, because actual value similarities are large, and because perceived value differences are often assumed but rarely corrected in everyday life. Future work may benefit from comparing or combining a value-based approach with other approaches. For instance, interventions could combine correcting misperceptions of ideological polarization (e.g., values) and support for anti-democratic attitudes (e.g., support for partisan violence; Mernyk et al., 2022), in light of evidence that improvements in ideological and affective polarization do not necessarily translate into effects on anti-democratic tendencies (e.g., Voelkel et al., 2023a). Future research could also test approaches that not only correct misperceptions of values but also of how these values are expressed in attitudes or behaviors (i.e., value instantiations; Maio, 2010) for a more comprehensive approach that simultaneously targets group differences at abstract and concrete levels.
Another important question for future research is the longevity of the effects. Past work has found that the effects can persist for several weeks after the intervention (Mernyk et al., 2022; but see: Lees & Cikara, 2020; Syropoulos & Leidner, 2023) but also that the effects are susceptible to introducing doubt (e.g., alluding to the uncertainty of science) or conflicting information (e.g., on differences in other variables; Druckman, 2023). Future work could test whether the present approach shows more enduring effects because perceptions of value similarities should remain personally meaningful and relevant across situations, and because the overlapping distribution method could build resistance against conflicting information that focuses exclusively on extremes and exaggerated group differences. Nevertheless, it should be acknowledged that information from a one-shot manipulation will not always be remembered and brought to mind in relevant situations, and we hence expect that repeated exposure paired with more explicit guidance (e.g., how to spot and correct for exaggerated differences) may be most effective.
The present research goes beyond past work in important ways. It presented information comparing real responses from U.S. Republicans and Democrats in graphical displays of overlapping distributions rather than presenting outgroup averages. In fact, similarity conditions in past work showing group averages to correct misperceptions of differences (Ahler, 2014) may be most comparable to our barplot conditions, suggesting that our intervention provides benefits beyond past work. Moreover, our approach not only provides a rich and visual intervention to correct people’s misperceptions of group differences, but this approach in our view also demonstrates an improved way of communicating group comparisons to the public. The social sciences and media may benefit from (additionally) reporting the extent of overlap between groups when communicating findings on group comparisons to the public (Hanel et al., 2019; Hyde, 2005), helping to correct overblown perceptions of polarization with positive effects for society.
Author
Topic: AmeriPol thread (Read 4150750 times)