Asymmetric affective polarization regarding COVID-19 vaccination in six European countries

While recent research has shown that supporters and opponents of COVID-19 vaccination have polarizing political attitudes and beliefs, we lack a thorough understanding of how these two groups think about each other. To investigate the feelings and stereotypes between supporters and opponents of COVID-19 vaccination, this study draws on cross-sectional survey data from six European countries (France, Germany, Italy, Spain, Switzerland, and the United Kingdom), collected between January and March 2022 (n = 6379). Our findings indicate an opinion-based affective polarization between supporters and opponents of COVID-19 vaccination. Both groups not only adopt different positions on the issue but also display dynamics of in-group favoritism and out-group hostility. Most notably, our assessment of thermometer scores and character trait ratings shows that this affective polarization is asymmetric, as it is stronger among the pro-vaccination group. Our findings are critical to the control of infectious diseases because affective polarization has been shown to influence health behaviors such as compliance with government policies. The issue is even more pressing as globalization boosts the threat of pandemic emergence and accelerates the global transmission of diseases.


Background
Affective polarization, understood as emotional attachment to in-group partisans and hostility toward out-group partisans, has become an important component of modern democracies 8,9 .Rooted in social identity theory 31 , the concept of affective polarization has long been studied with a focus on partisanship 8,32 .However, the identification and subsequent affective evaluation of in-and out-groups can also be based on identification with a common opinion rather than a party label 2,15,[33][34][35][36] .Particularly after dramatic events such as natural disasters, opinionbased affiliations tend to form around controversial issues that prompt diametrically opposed groups regarding the preservation of the status quo 36 .This is especially true for issues that cross-cut traditional lines of social categories or partisan conflict.More than just a tally of rational considerations 37 , this form of polarization rests on the assumption that people define their opinions as part of their identity, i.e., they "need to define themselves in terms of their opinion group membership in the same way that they would any other meaningful social group" 15 .Such opinion-based social identities contribute to the self-esteem of group members and thus inevitably lead to a tendency for positive in-group evaluation and a clear distinction from others outside the group 37 .
According to social identity theory, three mental processes are required for the formation of a social identity 31 .First, social categorization, in which we distinguish between "us" and "them".Second, social identification, in which we adopt a social identity based on our group membership, implying a sense of shared identity and belonging.Third, the social comparison of our in-group with an out-group, which in the context of group competition generates not only positive feelings toward the in-group but also negative feelings toward the out-group and thus conflictive intergroup relations, which in turn reinforce in-group identification 31 .
In this study, we focus on the formation of opinion-based affective polarization regarding COVID-19 vaccination.To illustrate this, Fig. 1 provides a conceptual model with two fictional characters, Anna and Peter,  www.nature.com/scientificreports/their opinions on COVID-19 vaccination, and their feelings toward supporters and opponents of this specific vaccine.While Peter is opposed to COVID-19 vaccination, Anna is in favor of it (ideological polarization, upper part of Fig. 1).Based on their respective ideological position with respect to the issue, Anna and Peter categorize themselves as pro-COVID-19 vaccine and anti-COVID-19 vaccine, respectively.This categorization is followed by a process of identification in which both begin to develop an opinionbased identity with their respective group.In other words, they start to see themselves in terms of their group's characteristics and incorporate its norms, values, and behaviors 31 .Thus, Anna sees herself as a member of the pro-COVID-19 vaccine camp and identifies with others from that camp, while Peter sees himself as a member of the anti-COVID-19 vaccine camp and connects with others from that camp (middle part of Fig. 1).
Having categorized and identified with one group, Anna and Peter begin to compare their group to the other.In a context of group competition, such comparisons are biased in favor of one's in-group in order to maintain self-esteem, which often leads to negative feelings toward the out-group 16,17,31 .The end result therefore is an affective bias between the groups that stems from the mental processes of categorization, identification, and comparison 16 .
Based on these considerations, we expect opinion-based group formation and affective polarization regarding COVID-19 vaccination.The debate about the benefits of vaccination and vaccine mandates was highly politicized during the COVID-19 pandemic.Previous research also suggests that divisions based on vaccination status emerged during the pandemic [1][2][3] .Thus, we consider it likely that the issue of COVID-19 triggered opinion-based group identification, as it was highly salient at the time and prompted individuals to take sides during this watershed public health crisis 16 .In other words, we expect pro-vaccination individuals to identify with others who share their views and to feel positive about other pro-vaccination individuals and negative about anti-vaccination individuals, while we expect the opposite for anti-vaccination individuals.Thus, if our arguments are supported, our data would show that Anna, as a pro-COVID-19 vaccination individual, would feel positive toward other pro-COVID-19 vaccination individuals and negative toward anti-COVID-19 vaccination individuals.The opposite pattern should be observed for Peter (lower part of Fig. 1).
While the key insights of social identity theory make us suggest that both groups would express affective polarization in similar ways, the perceived salience of the out-group is crucial for this intergroup affect 31 .In this regard, recent research indicates that both groups will differ in their expressed levels of intergroup affect 1,2,16 .In our case, one might expect higher levels of intergroup affect among vaccination supporters because they view immunization as a social contract to achieve collective security 5,38 .Consequently, vaccination opponents break this contract and act as free riders and norm violators 5,38 .In this vein, one might assume high levels of affective polarization among vaccination supporters because the out-group (vaccination opponents) is considered a rulebreaker, norm violator, and potentially threatening.
Conversely, opponents of the vaccine may focus their antipathy not on supporters of the vaccine but on political actors, such as the government.In their view, the norm violators are government actors who promote a vaccine that is seemingly unnecessary given the supposed controllability of the pandemic.In some countries, anti-vaccination groups have frequently taken to the streets, sometimes violently, to voice their discontent with the restrictions on their individual freedom imposed by lockdowns or vaccination policies 39 .However, these protests were often directed against the government.The regression analyses in the supplementary material, section B (Table 11) show that in all six countries studied, anti-vaccination individuals were significantly more distrustful of the government than pro-vaccination individuals were.Consequently, the negative affect toward vaccination supporters could be lower.Thus, the difference in salience of the opposing group for pro-and anti-vaccination individuals might have important implications for the level of intergroup affect.As a result, one might expect both in-group attachment and out-group dislike to be higher among the pro-vaccination group than among the anti-vaccination group.Referring to our conceptual model, Peter should be less affectively polarized than Anna.
The left panel in Fig. 2 shows the mean levels of support for COVID-19 vaccination across our six countries.Generally, support is very high with mean levels of 7 on the scale from 0 to 10.The lowest level of support is found in France and the highest in the UK and Spain.The violin plots in the right panel of Fig. 1 again illustrate this finding as the distribution is skewed in favor of COVID-19 vaccination.However, the distribution shows that a consistent minority opposes COVID-19 vaccination in all six countries.That is, the issue has triggered the formation of two groups: one for and one against it.
In a next step, we applied the most widely used measurement approach for affective polarization, the feeling thermometer 8,40 , in which respondents are asked to rate their feelings about a specific subject on a temperature scale 9,40 .First, we dichotomized the variable on the position regarding COVID-19 vaccination and classified respondents who indicated a value from 0 to 4 as opponents of COVID-19 vaccination (anti-vaccination group) and those who indicated a value from 6 to 10 as supporters of COVID-19 vaccination (pro-vaccination group).We excluded respondents who indicated a value of 5 as they were neutral on the issue.Second, we asked respondents of the respective groups to rate their feelings toward a) supporters of COVID-19 vaccination and b) opponents of COVID-19 vaccination on a scale from − 5 ("very cold and negative") to + 5 ("very warm and positive").We transformed this scale to range from 0 to 10 and subsequently used the absolute difference between Looking at the thermometer scores for COVID-19 vaccination opponents, we see a mirror image.Opponents feel relatively positive and warm toward other opponents ((M) Opp FRA = 6.55, (M) Opp GER = 6.73, (M) Opp ITA = 6.77, (M) Opp SPA = 5.78, (M) Opp SWI = 6.85, (M) Opp UK = 5.96).Yet, these positive in-group feelings are comparatively lower among opponents than among supporters.It seems that there is less group cohesion among opponents than among supporters of COVID-19 vaccination.Furthermore, supporters express very cold and negative feelings toward opponents, as expected ((M) Supp FRA = 2.46, (M) Supp GER = 2.07, (M) Supp ITA = 2.27, (M) Supp SPA = 1.76, (M) Supp SWI = 2.73, (M) Supp UK = 2.32).All differences between supporters and opponents are statistically significant at the 95% level.
Figure 4 reveals the absolute difference between both thermometer ratings, separated by group and country.Supporters show a relatively high average difference in feelings toward the in-group and the out-group ((M) Supp FRA = 6.63, (M) Supp GER = 7.17, (M) Supp ITA = 7.01, (M) Supp SPA = 7.35, (M) Supp SWI = 6.46, (M) Supp UK = 6.86).Opponents also show a difference in feelings for their in-and out-group, but this difference is less pronounced ((M) Opp FRA = 2.65, (M) Opp GER = 2.95, (M) Opp ITA = 3.18, (M) Opp SPA = 2.93, (M) Opp SWI = 2.87, (M) Opp UK = 3.02).All differences in affective polarization are significant at the 95% level.The highest levels of affective polarization are found in Germany and Spain among supporters and in Italy and the United Kingdom among opponents.Both groups express affective polarization regarding COVID-19 vaccination, but it is stronger among the pro-than among the anti-vaccination group.As expected, affective polarization is asymmetric, implying that both in-group attachment and out-group dislike are stronger among the pro-vaccination group than among the anti-vaccination group.

B. Measuring opinion-based affective polarization regarding COVID-19 vaccination using character trait ratings
To further evaluate our findings, we used a second common measure of affective polarization: character trait ratings 14,17,40 .Here, respondents from the pro-vaccination camp and the anti-vaccination camp (as coded above) www.nature.com/scientificreports/are asked to rate various character traits of the two different groups.Although trait ratings are a typical measure of affective polarization, they reflect more than just negative affect but also shed light on the perceived stereotypical appearance of a group 9 .In this vein, these trait ratings allow us to identify whether respondents assign negative or positive characteristics to their respective in-group and out-group.Research has shown that trait ratings and thermometer scores, although conceptually somewhat distinct, correlate fairly well with each other and show little systematic differences 40 .In our full sample, the affective polarization scores for both measures correlate at (r) Sample = 0.62 (r) Sample FRA = 0. 61, (r) Sample GER = 0.64, (r) Sample ITA = 0.67, (r) Sample SPA = 0.57, (r) Sample SWI = 0.59, (r) Sample UK = 0.65).
In our study, we asked respondents to rate the extent to which two positive character traits (openness to compromise and critical thinking) and two negative character traits (selfishness and narrow-mindedness) apply to a) supporters of COVID-19 vaccination and b) opponents of COVID-19 vaccination on a scale from 1 ("does not apply at all") to 5 ("fully applies").In addition to the assigned values, we also calculated the absolute differences between the scores assigned to the in-and the out-group for each trait.Subsequently, we combined these differences into an additive score for affective polarization ((M) Sample FRA = 1.72, (M) Sample GER = 2.08, (M) Sample ITA = 1.92, (M) Sample SPA = 1.99, (M) Sample SWI = 1.87, (M) Sample UK = 2.01).
This alternative measure reveals a similar picture of affective polarization regarding COVID-19 vaccination as the feeling thermometer.Figure 5 is analogous to Fig. 3 and shows the character trait ratings for supporters and opponents of COVID-19 vaccination by group and country.For the sake of readability, we combine the two positive and negative traits each (see supplementary material, section D, Fig. 1 for the individual character trait ratings).The upper panel of Fig. 5 shows the ratings of the two negative traits combined.As we can see in the upper left panel, supporters do not believe that other supporters are selfish and narrow-minded ((M) Supp FRA = 1.87, (M) Supp GER = 1.77, (M) Supp ITA = 2.15, (M) Supp SPA = 2.16, (M) Supp SWI = 2.00, (M) Supp UK = 1.68).Conversely, opponents tend to assign these negative traits to supporters ((M) Opp FRA = 2.63, (M) Opp GER = 3.00, (M) Opp ITA = 3.33, (M) Opp SPA = 2.79, (M) Opp SWI = 3.04, (M) Opp UK = 2.81).The opposite picture emerges when we look at the upper right panel: Consistent with in-group favoritism, opponents do not believe that other opponents are selfish or narrow-minded ((M) Opp FRA = 1.87, (M) Opp GER = 1.94, (M) Opp ITA = 2.08, (M) Opp SPA = 2.49, (M) Opp SWI = 1.98, (M) Opp UK = 2.23).Supporters, however, believe that opponents are selfish and narrowminded ((M) Supp FRA = 3.83, (M) Supp GER = 4.10, (M) Supp ITA = 4.08, (M) Supp SPA = 3.94, (M) Supp SWI = 3.96, (M) Supp UK = 3.91).All differences are significant at the 95% level.3 shows the mean thermometer ratings of feelings toward vaccination supporters and opponents separated by group and country, with 95% confidence intervals.Reading example for France in the left panel: In France, vaccination opponents have an average thermometer rating of feelings toward vaccination supporters of 6.15, while vaccination supporters have an average thermometer rating of feelings toward vaccination supporters of 7.04 on a scale of 0-10.The difference is statistically significant.
Although less pronounced, Fig. 5 provides further evidence that supporters and opponents of COVID-19 vaccination tend to view their in-group positively and their out-group negatively.Figure 6, which shows the absolute difference in character trait ratings by group and country, corroborates these observations.While both groups show a difference in ascribed character traits toward the in-group and the out-group, the average difference is again slightly but statistically significantly (at the 95% level) greater among vaccination supporters ((M) Supp FRA = 2.04, (M) Supp GER = 2.31, (M) Supp ITA = 2.09, (M) Supp SPA = 2.18, (M) Supp SWI = 2.03, (M) Supp UK = 2. Overall, our data suggest a divide around people's opinions on COVID-19 vaccination.Supporters and opponents dislike each other and tend to attribute positive traits to their in-group and negative traits to the respective out-group.Thus, our data indicates the presence of opinion-based affective polarization regarding COVID-19 vaccination in six European democracies in early 2022.Importantly, however, this affective polarization is asymmetric, as the pro-vaccination group tends to be more polarized than the anti-vaccination group.Naturally, this finding raises the question of potential correlates of this form of polarization.As a first step in this direction, exploratory analyses reported in the supplementary material, section E, Figs.2-4 show that older age, lower social trust, higher levels of conscientiousness, and a general support for COVID-19 vaccination are associated with higher levels of opinion-based affective polarization regarding COVID-19 vaccination.

Discussion
In this article, we shed light on the social fabric during the COVID-19 pandemic.Specifically, we extend research on affective polarization by focusing on opinion-based rather than partisan affective polarization.While partisan affective polarization has been described as a major challenge to democracy, partisan identities are not the  4 shows the mean level of affective polarization (thermometer measure) by group and country, with 95% confidence intervals.Reading example: In France, vaccination opponents show an average affective polarization of 2.65 and vaccination supporters of 6.63 on a scale from 0 to 10.The difference is statistically significant.only political identities that matter for intergroup hostility.Using COVID-19 vaccination as a prime example, we show that this issue formed the basis of an affective divide in the population in six European democracies in 2022.Applying both thermometer scores and trait ratings reflecting the perceived stereotypical appearance of a group, we observe an asymmetric affective polarization regarding COVID-19 vaccination in all six countries.While both supporters and opponents of COVID-19 vaccination show affective polarization, this polarization is higher among the pro-vaccination group 1,2 .
This report sets the stage for further research on opinion-based affective polarization that goes beyond the limitations of this study, which we address in the following.First, we operate with cross-sectional survey data and correlational analyses that are limited to a specific point in time.Experimental and longitudinal studies of group formation around COVID-19 vaccination would be a promising next step, particularly to see whether this division persists after the issue has faded in salience.Second, COVID-19 vaccination is a specific issue, raising the question of what other issues might serve as the basis for opinion-based group identities and subsequent affective polarization 15,16,33 .Third, although the finding of asymmetric affective polarization is consistent across the countries studied, some individual character traits show less consistency than others in the individual countries.This applies, for example, to the character trait of critical thinking in Italy and France.Furthermore, our analyses are limited to the character traits that have been examined in previous research on affective polarization.Future research is advised to expand the inventory and explore other characteristics such as honesty or trustworthiness 9 .In general, we must acknowledge that the expression of stereotypes seems to be less prevalent than direct dislike between the groups, as evidenced by the thermometer scores.
In addition, future studies could evaluate additional factors that may contribute to affective polarization, such as a tendency toward analytical thinking, different worldviews, and ideological orientations (e.g., hierarchy-egalitarianism and individualism-communitarianism).In addition, statistical techniques such as latent profile or class analysis might provide more insight into which groups in society express higher levels of dislike and stereotypical thinking about their out-groups.Fourth, future research could examine the consequences of opinion-based affective polarization and compare them to the presumed consequences of partisan affective polarization, particularly in terms of democratic norms, social cohesion, and health behavior 18,19,22,23 .
Despite these limitations, our findings have important policy implications regarding the challenges posed by infectious diseases.First, high levels of opinion-based affective polarization undermine efforts to implement public health measures and limit the role civil society can play in overcoming a pandemic.In-group favorability and out-group hostility put a strain on democratic societies, which in times of pandemic threat may undermine  5 shows the mean perceived character traits for vaccination supporters and opponents separated by group and country, with 95% confidence intervals.For example, in the top left-hand panel for France: In France, vaccination opponents perceive vaccination supporters to have negative traits with an average of 2.63 while vaccination supporters perceive vaccination supporters to have negative traits with an average of 1.87 on a scale of 1-5.The difference is statistically significant.a unified and coherent public response to contain the spread of infectious diseases.Second, affective polarization may cause individuals to selectively engage with information that aligns with their pre-existing beliefs, further reinforcing misperceptions [41][42][43][44][45] .Such misperceptions often contradict evidence-based public information and hamper public responses to infectious diseases.Third, social cohesion suffers from affective polarization and the division into an "us-versus-them" mentality.Yet, public health crises require a collective response, which is threatened by affective divisions that polarize civil society.Addressing affective polarization is a difficult task that will require a refinement of traditional public health strategies.Intergroup contact, social outreach, and the creation and communication of common overarching identities may prove useful in reducing affective polarization and fostering a more unified response to public health threats in the future [46][47][48][49] .

Methods
The data were collected between January 25 and March 8, 2022, in France, Germany, Italy, Spain, Switzerland, and the United Kingdom.The survey was conducted by the German survey company SurveyEngine, using quotasampling regarding gender, age, and education (and language in the case of Switzerland) yielding around 1000 respondents per country.During the data collection, the pandemic was still very present due to the emergence of the Omicron variant.More importantly, although vaccines were widely available, there were heated debates about the value and consequences of COVID-19 vaccination.

Ethical approval
Ethical clearance for the study was obtained from the Ethics Committee of the Faculty of Business Administration, Economics and Social Sciences of the University of Bern with the project number 092020.Participants provided informed consent and received a small compensation from the survey company for their participation.

Figure 1 .
Figure 1.Psychological processes of ideological and affective polarization regarding COVID-19 vaccination based on social identity theory (conceptual model).

Figure 3 .
Figure 3. Thermometer ratings of feelings toward COVID-19 vaccination supporters and opponents by group and country.Notes: Figure3shows the mean thermometer ratings of feelings toward vaccination supporters and opponents separated by group and country, with 95% confidence intervals.Reading example for France in the left panel: In France, vaccination opponents have an average thermometer rating of feelings toward vaccination supporters of 6.15, while vaccination supporters have an average thermometer rating of feelings toward vaccination supporters of 7.04 on a scale of 0-10.The difference is statistically significant.

Figure 4 .
Figure 4. Affective polarization of COVID-19 vaccination supporters and opponents.Notes: Figure4shows the mean level of affective polarization (thermometer measure) by group and country, with 95% confidence intervals.Reading example: In France, vaccination opponents show an average affective polarization of 2.65 and vaccination supporters of 6.63 on a scale from 0 to 10.The difference is statistically significant.

Figure 5 .
Figure 5. Perceived character traits of COVID-19 vaccination supporters and opponents by group and country.Notes: Figure5shows the mean perceived character traits for vaccination supporters and opponents separated by group and country, with 95% confidence intervals.For example, in the top left-hand panel for France: In France, vaccination opponents perceive vaccination supporters to have negative traits with an average of 2.63 while vaccination supporters perceive vaccination supporters to have negative traits with an average of 1.87 on a scale of 1-5.The difference is statistically significant.

Figure 6 .
Figure 6.Affective polarization among COVID-19 vaccination supporters and opponents.Notes: Figure6shows the mean level of affective polarization (according to the character trait ratings) by group and country, with 95% confidence intervals.Reading example: In France, vaccination opponents have an average affective polarization of 1.25, while vaccination supporters have an affective polarization of 2.04 on a scale from 1 to 5. The difference is statistically significant.