This is an open access article distributed under the terms of the Creative Commons
Attribution License (Whilst Democrats and Republicans in the USA have had similar views regarding some aspects of the coronavirus pandemic, there have been significant points of difference. Polling from the Pew Research Center (2020a) suggests that compared to Republicans, Democrats have taken the pandemic more seriously, and have been more likely to endorse steps to stop it from spreading. This finding is corroborated by observational data, for example mobile phone tracking data showing less evidence of social distancing in predominantly Republican areas (Allcott, Boxell, Conway, Gentzkow, Thaler, & Yang, 2020). There is ample evidence that political polarization affects attitudes, judgments and behaviors (Van Bavel & Pereira, 2018), and a polarized response to coronavirus has been particularly blatant in the USA (Van Bavel et al., 2020). This provides a unique opportunity to study how party ties can influence responses to a crisis. In the current research, we explored Republicans’ and Democrats’ responses to the coronavirus pandemic and the extent to which people are guided by their party’s position, potentially at the expense of their own and others’ personal safety.
Political polarization has increased dramatically in the USA in recent years (Pew Research Center, 2019), and this is likely to continue as people increasingly avoid ideologically-inconsistent messages and consume information within like-minded “echo chambers” (Flaxman, Goel, & Rao, 2016). In such echo chambers, facts often take second place to misinformation such as fake news (Lazer et al., 2018) and conspiracy theories (Douglas, Sutton, & Cichocka, 2017) that reinforce political biases and divisions. Political partisanship affects beliefs in a range of areas such as social attitudes (Petersen, Skov, Serritzlew, & Ramsøy, 2013), support for scientists and scientific findings (Kam, 2005; Kahan, Jenkins-Smith, & Braman, 2011), and support for social policy decisions (Cohen, 2003). It can even skew people’s memories for important political and social events (Frenda, Knowles, Saletan, & Loftus, 2013). Van Bavel and Pereira (2018) argue that “the tribal nature of the human mind leads people to value party dogma over truth” (p. 214), so much so that they often abandon their own values and beliefs in favor of party loyalty.
We consider if Republicans’ and Democrats’ responses to the coronavirus pandemic reflect uncritical acceptance of party cues, or instead a more active process of conformity to perceived party interests. In identity-based groups such as political parties, people are attracted to the group’s goals, activities and other defining features, and are strongly aware of, and comply with, group practices (Prentice, Miller, & Lightdale, 1994). People derive a sense of personal self-esteem and pride from being members of groups and adopting their interests (Tajfel, 1979). Displaying loyalty to one’s political party and actively serving one’s party’s interests can therefore be a way of maintaining a positive sense of social identity. That is, by doing what they think their political party wants them to do (and avoiding behaviors that appear to go against their party’s wishes), people might be able to affirm and express their identities. However, this conformity might have personal and social costs. Group conformity has been linked to negative outcomes such as bullying (Smith, 2016), prejudice (Nesdale, Maass, Durkin, & Griffiths, 2005), and unhealthy behaviors (Templeton, Stanton, & Zaki, 2016). Conformity to a group can also fuel group polarization and further cement negative attitudes (Koudenburg, Greijdanus, & Scheepers, 2019).
The current research is concerned with the sense in which we can say that Republicans’ and Democrats’ different attitudes, intentions, and behaviors regarding coronavirus are partisan. We argue that the motivation to think and act in ways that further party interests is straightforwardly partisan, consistent with psychological analyses of the construct (e.g., Van Bavel & Pereira, 2018). This motivated partisanship may be responsible for the gap between Republicans and Democrats, but other factors may be responsible. For example, compared to Democrats, Republicans tend to be more sceptical about at least some domains of science (Rutjens, Heine, Sutton, & van Harreveld, 2018), and albeit with some exceptions, are generally more opposed to perceived state intrusions on individual liberties—especially intrusions that threaten economic growth (Campbell & Kay, 2014). Furthermore, Republicans and Democrats could easily have diverged because of their passive consumption of media, such as right or left-leaning TV networks and newspapers, that tout different messages about the seriousness of coronavirus (Kushner Gadarian, Goodman, & Pepinsky, 2020).
These differences between Republicans and Democrats—as well as, or instead of, their desire to toe the party line on coronavirus—may explain their diverging attitudes. In this research, we therefore seek evidence that their divergence arises from perceptions of party interests. This requires that compared to Democrats, Republicans perceive it to be relatively against their party’s wishes or interests to take coronavirus seriously. It also requires that their different attitudes, intentions and behaviors concerning coronavirus reflect, at least in part, this perceived difference in party interests. In three studies, we tested these predictions. In Studies 1 and 2 we focused on participants’ attitudes and intentions toward a set of preventive measures. In Study 3 we focused specifically on participants’ mask-wearing attitudes, intentions, and behaviors. Overall, we predicted that perceptions of party interests would shape people’s attitudes concerning coronavirus and both behavioral intentions and actual behaviors designed to prevent its spread.
Study 1
Study 1 was conducted in the early days of the pandemic, at the beginning of March 2020. Information about the virus was rapidly emerging, and many attitudes were divided on partisan lines. We asked Republican and Democrat participants to indicate their attitudes concerning the seriousness of the virus, attitudes toward preventive actions to stop the virus from spreading, and their intentions to take those actions. We also asked participants to rate which political party’s interests they felt are best served by framing coronavirus as a threatening disease. We first predicted that participants’ attitudes and intentions would be driven by their political party affiliation, such that compared to Democrats, Republicans would be less inclined to believe that coronavirus is a serious threat, and be less inclined toward preventive actions. We further predicted that this difference would be mediated by perceived party interests. That is, the differences between Republicans and Democrats in coronavirus attitudes and intentions would be mediated by the extent to which they perceive it is in their own party’s interests to frame the virus as a serious threat.
Method
Participants and Design
Four hundred participants1 (Mage = 35.16, SD = 13.43; 225 female, 165 male, 3 trans, 4 non-binary, 3 rather not say) were recruited from Prolific and paid a small fee. Participants were selected as either Democrat or Republican as declared on their Prolific profile. However, we also asked participants to declare their political party affiliation in the questionnaire, and 189 indicated that they identified as Democrat, 191 Republican, and 20 indicated ‘other’ affiliations including Libertarian, Independent and Green. We included only Democrat and Republican participants and the final number of participants was therefore 380 (Mage = 35.53, SD = 13.54; 212 female, 161 male, 1 trans, 4 non-binary, 2 rather not say). One participant reported no formal education, 16 had primary education, 152 secondary, 154 college (Bachelor’s degree) and 57 college (graduate degree).
Materials and Procedure
The study was conducted on March 5th and 6th 2020 via Qualtrics. Participants gave their informed consent and then completed three measures, each with four items, on a five-point scale from 1 (strongly disagree) to 5 (strongly agree). These measured perceived seriousness (e.g., “The coronavirus is very mild and no worse than the seasonal flu”, α = .62), attitudes toward preventive actions (e.g., “The advice about frequent hand-washing is excessive and unnecessary”, α = .58), and intentions to carry them out (e.g., “I will not wash my hands more frequently”, α = .55).2 All items were reverse-scored such that higher scores indicated more perceived threat, more favorable attitudes toward actions, and greater intentions to take those actions. Since the scales were not reliable by conventional standards, and were not clearly differentiated in exploratory factor analyses, we also calculated a composite of the 12 items to form an index of coronavirus attitudes (α = .80).
Perceived party interest was measured with two items asking participants to indicate which of the two parties appears to want people to believe coronavirus is a threatening disease, and which benefits from people believing this, both from 1 (Democrats) through 5 (neither) to 9 (Republicans). These items were only modestly correlated (r = .43, p < .001) and were therefore analyzed separately. We included a quality check: “Did you complete all questions truthfully, and should we use your responses?”, to which all participants responded yes. Finally, participants were debriefed, thanked, and paid.
Results and Discussion
Data and materials for all studies are available on the following Open Science Framework link: https://osf.io/xyk8w.3 Means, standard deviations, and the results of univariate analyses of variance (ANOVA) are presented in Table 1. Compared to Democrats, Republicans were less concerned about coronavirus (but not significantly so), had significantly less favorable attitudes toward actions to combat the virus, and significantly weaker intentions to carry out them out.
Table 1
Means (and Standard Deviations) for All Dependent Measures Based on Political Party Affiliation (Study 1)
| Dependent measure | Republican | Democrat | ANOVA |
|---|---|---|---|
| Seriousness | 3.40 (.86) | 3.55 (.78) | F(1, 378) = 3.32, p = .069, η2 = .009 |
| Actions | 3.83 (.68) | 4.09 (.56) | F(1, 378) = 17.17, p < .001, η2 = .043 |
| Intentions | 3.48 (.74) | 3.70 (.73) | F(1, 378) = 8.96, p = .003, η2 = .023 |
| Coronavirus attitudes | 3.56 (.65) | 3.78 (.56) | F(1, 378) = 11.64, p = .001, η2 = .030 |
| My party wants people to take coronavirus seriously | 3.63 (2.07) | 6.07 (1.39) | F(1, 378) = 219.67, p < .001, η2 = .368 |
| My party benefits from people taking coronavirus seriously | 3.31 (1.93) | 5.53 (2.07) | F(1, 378) = 151.81, p < .001, η2 = .287 |
Note. In this table, the perceived party interest items were reverse-coded for Democrats, so that they reflect each participant’s perception that their own party, compared to its opponents, wants and benefits politically from people believing that coronavirus is a serious disease. Analysis of raw scores against the scale midpoint of 5 revealed a consensus that taking coronavirus seriously favors the Democrats, (“party wants”, M = 3.43, SD = 1.88, t(379) = -16.30, p < .001; “party benefits”, M = 3.88, SD = 1.84, t(379) = -11.91, p < .001). Though this consensus was endorsed by both groups, all ps < .001), the perception that Democrats will benefit more was stronger for Republicans than Democrats (M = 3.32, SD = 1.93 and M = 4.44, SD = 1.55 respectively), t(362.25) = -6.28, p < .0001. The difference for the “party wants” variable was not significantly different between Republicans (M = 3.51, SD = 1.98), and Democrats (M = 3.34, SD = 1.78), t(378) = 2.63, p = .397.
We then tested models of the mechanisms that may be responsible for the different beliefs, attitudes, and intentions of Republicans and Democrats. Analysis of the two candidate mediators was conducted with Process (Hayes, 2022), using 95% confidence intervals and 5,000 resamples (Model 6). For both candidate mediators, we entered party allegiance as the independent variable, and (composite) coronavirus attitudes as the dependent measure. This revealed the predicted indirect effect via the perception that one’s own party wants people to take coronavirus seriously. However, mediation via the perception that one’s own party benefits from people taking coronavirus seriously was not significant (see Figure 1). The direct effect between party allegiance and (composite) coronavirus attitudes was not significant.
Figure 1
Unstandardized Regression Coefficients for Mediation Analysis (Model 6)
Note. The direct effect between political party affiliation and coronavirus attitudes is in italics and the two indirect effects appear above and below the direct effect. Boldface highlights a significant effect as determined by 95% confidence intervals that do not cross zero. Note that in each study, Republican was coded ‘1’ and Democrat was coded as ‘2’.
Similar results were obtained for each of the three originally intended subscales. For perceived seriousness, indirect effect “party wants” = .14 [.09, .18], indirect effect “party benefits” = .02 [-.03, .07] while the direct effect was in the reverse direction, indicating that after party interests were taken into account, Republicans perceived coronavirus as a more serious illness, direct effect = -.29 [-.49, -.09]; for actions, indirect effect “party wants” = .08 [.04, .11], indirect effect “party benefits” = .01 [-.03, .05], direct effect = .01 [-.16, .15]; and for intentions, indirect effect “party wants” = .10 [.05, .14], indirect effect “party benefits” = .02 [-.03, .07], direct effect = -.10 [-.28, .08].
Study 1 therefore suggests that perceptions of party interests may influence people’s attitudes concerning coronavirus and behaviors designed to prevent its spread. Compared to Democrats, Republicans indicated that they believed the virus to be less of a serious threat, had less favorable attitudes toward preventive actions, and intended to take those actions less. These differences were mediated by the belief that their party does not want people view the virus as a serious illness.
Study 2
In Study 2, we sought to replicate the findings of Study 1 a few days later, when coronavirus was receiving more media and political attention in the USA, whilst also examining the roles of participants’ political ideology (left-right) and strength of party identification which were not included in Study 1. In Study 2, we also attempted to manipulate partisan motivations by making the salience of the Democratic party’s political interests experimentally salient. Specifically, we included an experimental condition emphasizing that taking coronavirus seriously is in the Democratic party’s interests, and a control condition where no further information was included. We expected the relationships between political party affiliation and preventive attitudes and intentions to be stronger when the partisan motivation of the Democratic party, compared to the Republican party, was made salient. That is, compared to controls, Republicans in the experimental condition should believe coronavirus to be less serious, and be less inclined toward coronavirus preventive behaviors, compared to those in the control condition. On the other hand, Democrats should show the opposite pattern of attitudes and intentions.
Method
Participants and Design
Four hundred and fifty five participants (Mage = 33.52; SD = 12.75, 268 female, 178 male, 2 trans, 7 non-binary) were recruited from Prolific on the basis of their Democrat or Republican status on their Prolific profile. However, in our survey, 220 indicated that they identified as Democrat, 210 Republican, and 25 indicated ‘other’ affiliations including Independent, Libertarian and Green. Our final number of participants was therefore 430 (Mage = 33.48, SD = 12.82; 256 female, 167 male, 2 trans, 5 non-binary). Two reported no formal education, 14 primary, 147 secondary, 198 college (Bachelor’s degree) and 69 college (graduate degree). The design was a 2 (salience of partisanship: political statement vs. control) x 2 (political party affiliation: Republican vs. Democrat) between-groups design.
Materials and Procedure
The study was conducted on March 9th and 10th 2020 via Qualtrics. Participants were randomly allocated to the experimental or control condition. In the experimental condition, participants read a brief paragraph entitled “Politics and the coronavirus” with the Democrat and Republican icons alongside the title. The text read “There is widespread agreement that compared to the Republican party, the Democratic party appear to want people to take the disease more seriously. We want to understand what you think about the virus”. Participants in the control condition received the heading “The coronavirus” and the last sentence of the paragraph only. Participants then completed the scales of perceived seriousness (α = .61), attitudes toward preventive actions (α = .60), and intentions (α = .58) as in Study 1, and their responses were reverse-scored.4 As in Study 1, we calculated a composite of the 12 items to form an index of coronavirus attitudes (α = .80). We were conducting this research when the coronavirus situation was rapidly changing and proceeded with Study 2 before discovering the low scale reliabilities in Study 1. Ideally, we would have adapted these measures before going ahead with Study 2. Nevertheless, the composite scale was as reliable as in Study 1 and therefore provided a reliable dependent measure to test our hypotheses.
Again, participants were asked to indicate which of the two parties appears to want people to believe coronavirus is a serious disease, and which political party is most likely to benefit if people believe this, both from 1 (Democrats) through 5 (neither) to 9 (Republicans). These items were again only modestly correlated (r = .26, p < .001) so we analyzed them separately. Participants also rated their identification with their party from 1 (not at all) to 7 (very much) and their political orientation (1 = extremely left-wing, 2 = moderately left-wing, 3 = somewhat left-wing, 4 = center, 5 = somewhat right-wing, 6 = moderately right-wing, and 7 = extremely right-wing). Participants were then debriefed, thanked, and paid.
Results and Discussion
We initially conducted two-way univariate ANOVAs for all dependent measures as a function of political party affiliation and experimental condition. However, it was clear that the manipulation of partisanship salience was not successful. A 2 x 2 ANOVA of party allegiance and salience condition on perceptions of party interest showed no main effect of the manipulation, F(1, 426) = 0.07, p = .788, η2 = .000, and crucially, no interaction with party allegiance, F(1, 426) = 1.55, p = .214, η2 = .004. Of the 430 valid participants, 178 failed the attention check that probed whether they noticed the manipulation, and analyzing results without these participants did not affect results. Neither did the manipulation have any main or interactive effects (ps > .207). Thus, we collapsed across conditions of the manipulation and report the results of one-way univariate ANOVAs of the effect of party identification, to examine if results replicate those of Study 1. Means and standard deviations are presented in Table 2. For each dependent variable, there was a main effect of political party affiliation, as in Study 1.
Table 2
Means (and Standard Deviations) for All Dependent Measures Based on Political Party Affiliation (Study 2)
| Dependent measure | Republican | Democrat | ANOVA |
|---|---|---|---|
| Seriousness | 3.23 (.83) | 3.63 (.78) | F(1, 428) = 27.44, p < .001, η2 = .060 |
| Actions | 3.75 (.69) | 4.21 (.58) | F(1, 428) = 53.96, p < .001, η2 = .112 |
| Intentions | 3.43 (.76) | 3.91 (.68) | F(1, 428) = 47.47, p < .001, η2 = .100 |
| Coronavirus attitudes | 3.47 (.47) | 3.92 (.55) | F(1, 428) = 59.82, p < .001, η2 = .123 |
| My party wants people to take coronavirus seriously | 3.59 (2.02) | 6.93 (1.75) | F(1, 428) = 335.39, p < .001, η2 = .439 |
| My party benefits from people taking coronavirus seriously | 3.11 (1.92) | 5.37 (1.32) | F(1, 428) = 203.35, p < .001, η2 = .322 |
Note. In this table, the perceived party interest items were reverse-coded for Democrats, so that they reflect each participant’s perception that their own party, compared to its opponents, wants and benefits politically from people believing that coronavirus is a serious disease. Analysis of raw scores against the scale midpoint of 5 revealed a consensus that taking coronavirus seriously favors the Democrats, (“party wants”, M = 3.34, SD = 1.90, t(429) = -18.61, p < .001; “party benefits”, M = 3.91, SD = 1.80, t(429) = -12.94, p < .001). Though this consensus was endorsed by both groups, all ps < .001), the perception that Democrats will benefit more was stronger for Republicans than Democrats (M = 3.11, SD = 1.32 and M = 4.63, SD = 1.92 respectively), t(368.99) = -9.62, p < .0001. The difference for the “party wants” variable was significant and endorsed more by Democrats (M = 3.07, SD = 1.75) than Republicans (M = 3.59, SD = 2.02), t(429) = 3.71, p = .005.
As in Study 1, we then tested models of the mechanisms that may be responsible for the different beliefs, attitudes, and intentions of Republicans and Democrats. This revealed the predicted indirect effect via the perception that one’s own party wants people to take coronavirus seriously. However, mediation via the perception that one’s own party benefits from people taking coronavirus seriously was again not significant (see Figure 2). The direct effect between party allegiance and (composite) coronavirus attitudes was, however, significant.
Figure 2
Unstandardized Regression Coefficients for Mediation Analysis (Model 6)
Note. The direct effect between political party affiliation and coronavirus attitudes is in italics and the two indirect effects appear above and below the direct effect. Boldface highlights a significant effect as determined by 95% confidence intervals that do not cross zero.
Similar results were obtained for each of the three originally intended subscales. For perceived seriousness, indirect effect “party wants” = .08 [.04, .12], indirect effect “party benefits” = .01 [-.04, .07], direct effect = .11 [-.10, .32]; for actions, indirect effect “party wants” = .10 [.07, .13], indirect effect “party benefits” = -.01 [-.05, .03], direct effect = .15 [-.01, .31]. For intentions, indirect effect “party wants” = .27 [.14, .40], indirect effect “party benefits” = .00 [-.06, .06], but this time the direct effect was also significant = 21 [.02, .40].
All of the ANOVA and indirect effects were the same when political ideology (left-right) and strength of identification were controlled for. We found no evidence that strength of identification moderated any of the effects. Therefore, despite an unsuccessful manipulation of partisan motivations, Study 2 further suggests that people’s perceptions of party interests may influence their attitudes concerning coronavirus, and their intentions to take preventive actions.
Study 3
Study 3 was designed to provide further evidence for the hypothesis that perceived party interests mediate the effect of party allegiance on responses to coronavirus. Instead of focusing on attitudes and intentions generally, we focused specifically on mask-wearing in public places. The study was conducted in June 2020, during which time mask-wearing was the subject of intense political tension in the USA. Compared to Democrats, Republicans were more opposed to the idea of wearing masks (Pew Research Center, 2020b). This difference therefore provided a good opportunity to further explore the relationships between political party affiliation, perceived party interests, and coronavirus responses.
Furthermore, to address the limitations of the measures used in Studies 1 and 2, we included separate measures of people’s attitudes towards wearing masks, their intentions to wear masks, and their reported mask-wearing behavior. As in Studies 1 and 2, we also measured perceived party interests. We therefore tested the hypotheses that Republicans would have less favorable attitudes toward wearing masks, would be less inclined to wear masks, and would indeed report wearing masks less often, mediated by the perceptions that it is against their party interests to have these attitudes, intentions, and to perform these behaviors.
Method
Participants and Design
As in Studies 1 and 2, participants (N = 401, Mage = 36.22; SD = 11.19, 205 female, 192 male, 3 non-binary or other gender identity, 1 rather not say) were recruited from Prolific on the basis of their Democrat or Republican status on their Prolific profile. However, in our survey, 199 indicated that they identified as Democrat, 194 Republican, and 8 indicated ‘other’ affiliations including Independent, Libertarian and Socialist. Our final number of participants was therefore 393 (Mage = 36.40, SD = 11.21; 202 female, 187 male, 3 non-binary or other gender identity, 1 rather not say). One reported no formal education, 9 primary, 89 secondary, 207 college (Bachelor’s degree) and 87 college (graduate degree).
Materials and Procedure
The study was conducted on 24th June 2020 via Qualtrics. Participants indicated their attitudes (opposition) toward wearing a mask by completing three questions (e.g., “Being required to wear a mask or other type of face covering is an infringement of my freedom of choice”, from 1 (strongly disagree) through 4 (neither agree nor disagree) to 7 (strongly agree), α = .82). Participants also indicated their intention to wear a mask with one item: “I intend to wear a mask or other type of face covering” from 1 (strongly disagree) through 4 (neither agree nor disagree) to 7 (strongly agree). They also indicated their own mask-wearing behavior with one item: “In public places where social distancing is difficult, do you wear a mask or other type of face covering? (1 = never 2 = very rarely, 3 = rarely, 4 = sometimes, 5 = often 6 = very often, 7 = always). Participants were given the option to choose “not applicable” if they do not leave the house. These participants were removed from the analysis. To measure perceived party interests, participants were asked to indicate which of the two parties appears to want people to believe coronavirus is a serious disease, from 1 (Democrats) through 5 (neither) to 9 (Republicans). We removed the “benefit” item which had not been a significant mediator in the two previous studies. Participants also rated their identification with their party from 1 (not at all) to 7 (very much).5 We included a quality check: “Did you complete all questions truthfully, and should we use your responses?”, to which all participants responded yes. They were then debriefed, thanked, and paid.
Results and Discussion
Means and standard deviations are presented in Table 3. For each dependent variable, there was a main effect of political party affiliation, replicating Studies 1 and 2 with the more specific coronavirus response of wearing masks in public places.
Table 3
Means (and Standard Deviations) for All Dependent Measures Based on Political Party Affiliation (Study 3)
| Dependent measure | Republican | Democrat | ANOVA |
|---|---|---|---|
| Negative attitudes | 3.34 (1.72) | 1.90 (1.07) | F(1, 391) = 99.13, p < .001, η2 = .202 |
| Intentions | 5.45 (1.85) | 6.54 (1.01) | F(1, 391) = 53.15, p < .001, η2 = .120 |
| Behavior | 5.30 (1.98) | 6.43 (1.09) | F(1, 377) = 47.15, p < .001, η2 = .110 |
| My party wants people to take coronavirus seriously | 4.23 (2.76) | 8.10 (2.89) | F(1, 391) = 318.89, p < .001, η2 = .449 |
Note. Analyses of raw scores showed that participants aligned with both parties (both ps < .001) agreed that the Democrats appear to want people to believe that coronavirus is a serious disease compared to Republicans, (M = 3.05, SD = 2.44), one-sample t(392) = -15.83, p < .001, and Democrats endorsed this more strongly (M = 1.90, SD = 1.30) than Republicans (M = 4.23, SD = 2.76), t(273.54) = 10.68, p < .001.
We again conducted mediation analyses in Process, using 95% confidence intervals and 5,000 resamples (Model 4). We found evidence for the hypothesis that perceived party interests mediated the effect of party allegiance on negative attitudes towards mask-wearing (Figure 3a). It also mediated the effect of party allegiance on intentions to wear masks (Figure 3b). Perceived own party interests also mediated the effect of party allegiance on reported mask-wearing behavior (Figure 3c).
Figure 3
Unstandardized Regression Coefficients for Mediation Analyses (Model 4)
Note. The direct effects between political party affiliation and the respective dependent variable are in italics and the indirect effects appear above direct effects. Boldface highlights a significant effect as determined by 95% confidence intervals that do not cross zero.
All of the significant ANOVA and indirect effects survived when strength of identification was controlled for. We also found no evidence that strength of identification moderated any of the effects. In summary, Republicans had less positive attitudes, were significantly less willing to wear masks, and reported wearing masks less, than Democrats. As in Studies 1 and 2, we found evidence that these differences were mediated by perceived party interests. In other words, Republicans were more opposed, less willing, and less likely to wear masks. This difference was mediated by the perception that it goes against their party’s interests to view the virus as a serious threat.
General Discussion
The current research suggests that partisan differences in willingness to protect oneself and others from coronavirus may at least in part reflect people’s active conformity to perceived party interests. There is a consensus that the Democrats want people to see coronavirus as a serious threat, and that the Democrats, compared to Republicans, are more motivated to ensure that people take protective measures. The current findings suggest that this perception of party interests may have shaped the tendency for Republicans to see coronavirus as less serious, and to be less favorable toward preventive behaviors, including wearing a mask in public places. The difference between Democrats and Republicans therefore seems to be partisan—it may emerge at least in part from an understanding and willingness to comply with the party’s perceived interests and what is at stake for them in the pandemic.
In light of a wealth of evidence that behaviors are not always reliably predicted by attitudes (e.g., Glasman & Albarracin, 2006), we find it particularly interesting that perceived party interests – in particular the perception that one’s own party wants people to treat coronavirus as a serious threat – mediated the effect of party allegiance on reported mask-wearing behavior. Research suggests that over the course of the pandemic, people have experienced several barriers to wearing masks such as concerns about infringement of personal freedoms (Lehmann & Lehmann, 2021), perceived threats to masculinity (Cassino & Besen-Cassino, 2020), and the perceived negative impact on social interactions (Carbon, 2020). It is therefore important to demonstrate that party differences in mask-wearing behavior appear to be predicted in part by perceived party interests. It suggests that Republicans may be consciously choosing not to wear masks because they believe it is not in their party’s interests to do so.
Limitations and Future Directions
Although we tested equal numbers of Republicans and Democrats, a limitation of the current research is that the samples are not nationally representative. Future research could therefore explore the links between partisanship and coronavirus responses across the US population, and also in different national settings. It is also worth noting that the data from Studies 1 and 2 were collected in early March 2020 and as the information and political framing of coronavirus changes, people’s response may also be changing. Although data for Study 3 were collected in June 2020 and mirror those of Studies 1 and 2, the political landscape has changed even further since then. Future research that builds longitudinal models of partisan effects over time would therefore be beneficial to extend on the current research and further illuminate the role of perceived political interests on responses to coronavirus and future crises.
Another limitation of the present research is that the findings are cross-sectional and we can therefore not infer causality. That is, we cannot conclude that political party allegiance causes individuals to reject health guidelines, and we likewise cannot conclude that perceived party interests cause changes in attitudes, intentions, or behaviors. Furthermore, we cannot rule out alternative explanations for our findings. For example, people’s attitudes and behaviors may be driving the belief that one’s party is in agreement with their personal attitudes. Although our theorizing has been derived from previous theory and research, we cannot rule out alternative perspectives without experimental or longitudinal studies.
Furthermore, we have only examined the mediating role of perceived party interests. There may be other important variables that are crucial in determining how political party affiliation is associated with political attitudes, intentions, and behaviors. In the current research, we accounted for strength of partisan identity, but future research should attempt to control for other variables. Finally, an important limitation of this work is that it measured perceived party interests using just two measures and these measures have not been used in the literature before. Also, findings were not consistent across these two measures. Specifically, the mediation paths were significant for perceptions of what the party wants but were not significant for perceptions of how the party benefits. This raises interesting theoretical questions about the differences between these two perceptions. For example, both may be types of motivated reasoning but quite different mechanisms. However, it also raises the concern that there may be alternative ways to measure perceived party interests that are not captured in the present studies.
Another important consideration for future research is to consider steps to break the partisan hold on people so that they are more likely to follow scientifically-based advice regarding coronavirus and future pandemics. Van Bavel and Pereira (2018) recognize the difficulty of doing so, but argue that fulfilling social identity needs through non-partisan means, motivating people to search for the truth, or activating other social identities and super-ordinate social identities, may effectively break the partisan bind.
Conclusion
The current research lifts the hood on partisan dynamics that have occurred during the coronavirus pandemic, and highlights the potentially detrimental consequences of a divided political landscape for important health-related decisions. It suggests that people might sometimes prefer partisan decisions despite appreciating the dangers they pose to their own health and the health of their community. On 8th April 2020, the Director General of the World Health Organization asked the then US president Donald Trump “please don’t politicize this virus”. The potentially catastrophic consequences of a politically divided view on coronavirus have been apparent since the beginning of the pandemic, and continue to damage efforts to eradicate the virus.