Introduction
Research on affective polarization has boomed in recent years. The origins of this thriving research agenda on affective polarization – that is, the gap in affect and sympathy towards supporters of different political parties – lie in the USA. The real-world triggers were the tumultuous political events of the past fifteen years, epitomized by the open inter-party hostility engendered by the election of Barack Obama and then Donald Trump, culminating (hopefully) in the 6 January storming of the Capitol. The academic trigger for research on affective polarization was a paper by Iyengar et al. (Reference Iyengar, Sood and Lelkes2012), which first highlighted that the gap between in-party and out-party feelings had been increasing steadily in the USA, and this mostly because out-partisan dislike had grown. A few years later, the influential work of Achen and Bartels (Reference Achen and Bartels2017) popularized the notion that groups and identities are the foundation of politics in the USA. Since then, the amount of research on the political identities, affective polarization and partisan hostility has been exceptional.
The majority of this research still focuses on the USA, but the last five years have seen researchers also turn to multi-party systems such as in Europe (Röllicke, Reference Röllicke2023). Footnote 1 But while in Europe the research agenda and the term ‘affective polarization’ are both new, studies looking at the phenomenon of out-party dislike and partisan animosity actually long predate this recent trend. In their landmark work, Almond and Verba (Reference Almond and Verba1963) ask now-standard questions concerning the perceived social distance of out-partisans in the UK, Germany and Italy, and this was also studied by Powell (Reference Powell1970) in his PhD dissertation on Austria (see also Engelmann and Schwartz (Reference Engelmann and Schwartz1974)). While Iyengar et al. (Reference Iyengar, Sood and Lelkes2012) cite and discuss this work, the long pedigree of this research is often forgotten, a fact also lamented by Schedler (Reference Schedler2023). Taking a further step back, the foundational work by Lipset and Rokkan (Reference Lipset, Rokkan, Lipset and Rokkan1967) is based on the notion of group conflict based around clear identities and social cleavages. The idea that groups and identities are important is not a recent contribution to understanding European politics; it is thus all the more important to be clear about what studying affective polarization adds to our knowledge.
This state-of-the-art review of research on affective polarization in Europe has three aims. First, I will describe how affective polarization has been conceptualized and measured in the types of multi-party systems prevalent in Europe. Second, I will summarize recent research findings, focusing on what studying Europe teaches about affective polarization and vice versa. Third, I will suggest that research on affective polarization in Europe faces four challenges that should be addressed by future research.
Affective polarization in Europe: conceptualization and measurement
Affective polarization refers to the extent to which supporters of political parties dislike and distrust one another (Druckman and Levendusky, Reference Druckman and Levendusky2019; Iyengar et al., Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019). More specifically, the term captures the distance between the sympathy individuals hold towards their in-party and the animosity they hold towards out-parties. As a ‘horizontal’ evaluation, capturing patterns of like and dislike at the citizen level, it is distinct from ‘vertical’ evaluations of partisan elites (Harteveld, Reference Harteveld2021a).
While in the USA two-party system, researchers can simply look at the gap between the in-party and the out-party to assess affective polarization, applying the concept to multi-party systems such as those predominant in Europe poses immediate challenges. First, positive party identification is generally in decline in Europe (Heath, Reference Heath, Fisher, Fieldhouse, Franklin, Gibson, Cantijoch and Wlezien2017), while multiple party identification is possible (e.g., Kekkonen et al., Reference Kekkonen, Suuronen, Kawecki and Strandberg2022). Complex party systems mean that Europeans are not faced with two partisan groups of roughly equal size, but with a sometimes dizzying array of unstable and ephemeral parties. In such contexts, affective polarization is best conceived of as either the average dislike of out-party supporters (compared to one’s sympathy for the most favoured party), or as the overall spread of affect across all partisans (Wagner, Reference Wagner2021). The diversity of party systems also means that it may make sense to focus on partisan animosity towards each party separately, so the extent to which their supporters are disliked or not (Gidron et al., Reference Gidron, Adams and Horne2023). A related question, further discussed below, is whether other political divisions provide the binary distinctions that party systems fail to provide in Europe. Hence, political identities and affective evaluations can be based on ideological identities (Oshri et al., Reference Oshri, Yair and Huddy2022; Bantel, Reference Bantel2023), issue preferences (Hrbková et al., Reference Hrbková, Voda and Havlk2023), Brexit stances (Hobolt et al., Reference Hobolt, Leeper and Tilley2021) or preferences for regional independence (Balcells and Kuo, Reference Balcells and Kuo2023).
Turning to questions of measurement, research on affective polarization can only make use of one type of question to examine historical patterns and developments in Europe over time: standard 0–10 like–dislike questions, where respondents are asked to state how much they like or dislike parties. This scale is the equivalent of the 0–100 warm–cold feeling thermometer used in the USA, most notably in the American National Election Study. The responses to this question provide information on how each individual relates to the party systems: respondents vary in how polarized their personal pattern of affect towards parties is. Aggregated to a higher level, these questions provide information on the mean level of affective polarization in a larger group – often the country.
The key advantage of these questions is that they form part of the standard repertoire of survey research on political attitudes and electoral behaviour. For example, they have been included in every round of the Comparative Study of Electoral Systems dataset (see, e.g., Ward and Tavits, Reference Ward and Tavits2019; Reiljan, Reference Reiljan2020; Wagner, Reference Wagner2021), which covers countries around the world, not just in Europe. Like–dislike scores have also been collected monthly in Germany since 1977 as part of the Politbarometer survey (Hudde, Reference Hudde2022; Harteveld and Wagner, Reference Harteveld and Wagner2023) and yearly in Sweden since 1986 as part of the SOM survey. Boxell et al. (Reference Boxell, Gentzkow and Shapiro2022) and Garzia et al. (Reference Garzia, Ferreira da Silva and Maye2023) provide useful overviews of the over-time availability of like–dislike scores or feeling thermometers across OECD countries.
In multi-party systems, three decisions need to be made when using like–dislike scores (Wagner, Reference Wagner2021). First, do citizens only have one in-party, or can they have several? Depending on the answer to this question, researchers should either use the mean like–dislike distance from the in-party or a type of standard deviation of like–dislike scores. Second, researchers need to decide whether to weigh parties by their size, so that, for example, disliked parties matter more if they are larger. Third, researchers need to decide whether affective polarization is only a relevant concept for those who have a positive partisan identity. Reiljan (Reference Reiljan2020), for example, restricts affective polarization to those who have an in-party, but if researchers want to assess overall patterns of partisan affect, this restriction is arguably not strictly necessary (Wagner, Reference Wagner2021). At the very least researchers need to decide whether they are interested in affective polarization among the whole electorate or only among partisans (Garzia et al., Reference Garzia, Ferreira da Silva and Maye2023).
One disadvantage of the predominant party-focused like–dislike questions is that they do not explicitly prompt affect towards partisans. Hence, by definition they arguably do not capture a core component of affective polarization, namely the fact that it relates to mutual affect of run-of-the-mill party supporters towards each other, rather than of voters towards party elites. Importantly, researchers have shown that the measures of like–dislike towards parties and towards partisans correlate closely. Gidron et al. (Reference Gidron, Sheffer and Mor2022), using data from Israel, find that the party-level like–dislike scale also captures sentiment towards party supporters and correlates with preferences for social distance and behavioural measures of discrimination. Using data from the Netherlands, Harteveld (Reference Harteveld2021a) shows that the individual-level correlation between like–dislike scores for parties and partisans is 0.66 (similar to the correlation reported for the USA in Iyengar et al., Reference Iyengar, Sood and Lelkes2012); the two scores thus correlate moderately but far from perfectly. To a certain extent, these findings are reassuring, and we should thus expect that existing results could largely be replicated even if a partisan like–dislike scale were used. Helpfully, more recent surveys have included like–dislike measures towards party supporters rather than parties, for example, the Spanish E-Dem survey or the Dutch LISS panel (Torcal et al., Reference Torcal, Santana, Carty and Comellas2020; Harteveld, Reference Harteveld2021a). It is likely that such measures will become more widely implemented in the future.
Related work has looked at like–dislike scores towards political leaders (Torcal and Comellas, Reference Torcal and Comellas2022; Reiljan et al., Reference Reiljan, Garzia, Ferreira Da Silva and Trechsel2023). Such scores are available over time in the Comparative Study of Electoral Systems and in the Comparative National Elections Project. Affective polarization scores can be calculated with these survey questions in the same way they can for political parties. In some contexts, such as presidential systems, leader-based scores may be more useful than party-based scores. In Brazil, for example, it was a leader – Jair Bolsonaro – who was one basis for broader societal polarization (Areal, Reference Areal2022). Usually, though, leader affective polarization is lower than party affective polarization (Reiljan et al., Reference Reiljan, Garzia, Ferreira Da Silva and Trechsel2023). Like the party-based scores, leader-based scores are unlikely to fully capture the affect between party supporters at a more horizontal level.
There has also been innovation concerning the measurement of partisan identities, a core component of theories of affective polarization (Huddy et al., Reference Huddy, Mason and Aarøe2015; West and Iyengar, Reference West and Iyengar2022). Thus, Huddy et al. (Reference Huddy, Mason and Aarøe2015) suggested a battery of items to capture positive partisan identity, and this has been adapted for multi-party systems by Bankert et al. (Reference Bankert, Huddy and Rosema2017) and Huddy et al. (Reference Huddy, Bankert and Davies2018). An interesting addition to this research is the development of a scale for negative partisan identities (Bankert, Reference Bankert2021; Areal, Reference Areal2022; Mayer and Russo, Reference Mayer and Russo2023). The concept of negative partisanship has been applied to European contexts using simpler measures in Mayer (Reference Mayer2017) and Meléndez and Kaltwasser (Reference Meléndez and Kaltwasser2021).
Alternative measures to like–dislike questions also include social distance questions (Bogardus, Reference Bogardus1933). These allow respondents to state whether they would be happy or unhappy to have party supporters as their colleague, neighbour or relative (Helbling and Jungkunz, Reference Helbling and Jungkunz2020; Knudsen, Reference Knudsen2021; Gidron et al., Reference Gidron, Sheffer and Mor2022). Kekkonen et al. (Reference Kekkonen, Suuronen, Kawecki and Strandberg2022) showed that social distance is consistently lower than simple out-party dislike. Trait ratings are another popular measure, as used most notably in Hobolt et al. (Reference Hobolt, Leeper and Tilley2021). In such questions, respondents assess supporters of different parties based on different positive and negative traits such as honesty or trustworthiness.
However, measures of group traits and social distance depart somewhat from a pure measurement of affective polarization (Röllicke, Reference Röllicke2023). Thus, social distance questions tell us not only about affect towards parties, but also the extent to which people care about spending time or talking to people who they dislike. Responses might thus also reflect personality characteristics or tendencies towards conflict avoidance (Ulbig and Funk, Reference Ulbig and Funk1999). Moreover, many people simply dislike those who talk about politics, irrespective of the views they hold or the party they support (Druckman and Levendusky, Reference Druckman and Levendusky2019; Druckman et al., Reference Druckman, Klar, Krupnikov, Levendusky and Ryan2022; Krupnikov and Ryan, Reference Krupnikov and Ryan2022). Similarly, trait ratings reflect more than just affect, capturing also the perceived stereotypical competence of a group (Cuddy et al., Reference Cuddy, Fiske and Glick2008) More practically, social distance and trait ratings are hard to aggregate into a summary affective polarization score and are perhaps best suited for measuring out-party hostility, a key component of affective polarization.
A more complex approach is to use survey experiments to assess affective polarization. Typically, voters might need to assess different hypothetical individuals in a vignette-based conjoint analysis (e.g., Helbling and Jungkunz, Reference Helbling and Jungkunz2020; Hrbková et al., Reference Hrbková, Voda and Havlk2023), while some research also uses economic games to capture the behavioural implications of partisan divides (e.g., Gidron et al., Reference Gidron, Sheffer and Mor2022). While these are very useful for measuring out-partisan affect and discrimination, they tend to provide aggregate assessments rather than information on each respondent.
What studying political conflict in Europe can tell us about affective polarization
Affective polarization has already been studied extensively (and arguably almost exhaustively) in the USA, so it is essential to ask what studying affective polarization in Europe – or multi-party systems in general – adds to scientific understanding, apart from additional cases. Of course, adding cases is itself important: one simple benefit of looking beyond the USA is to confirm that findings from that unique political system transfer to other settings. Thus, some studies (e.g., Algara and Zur, Reference Algara and Zur2023; Stoetzer et al., Reference Stoetzer, Munzert, Lowe, Çal, Gohdes, Helbling, Maxwell and Traunmueller2023) usefully include the USA as one of their cases. Moreover, at the most basic level, comparative research helps us to place individual cases – such as the USA – within a broader context. Research on affective polarization in the USA shows that partisan hostility seems to have reached alarming levels (Kalmoe and Mason, Reference Kalmoe and Mason2022), but is this a universal phenomenon?
Two findings stand out. First, affective polarization in the USA is by no means unusually high (Westwood et al., Reference Westwood, Iyengar, Walgrave, Leonisio, Miller and Strijbis2018; Gidron et al., Reference Gidron, Adams and Horne2020; Reiljan, Reference Reiljan2020; Knudsen, Reference Knudsen2021;Wagner, Reference Wagner2021; Garzia et al., Reference Garzia, Ferreira da Silva and Maye2023). Moreover, as in the USA, positive partisanship in Europe is not merely instrumental, but also is often strongly expressive, relating to deeply held identities (Huddy et al., Reference Huddy, Bankert and Davies2018). However, the USA may be characterized by particularly strong polarization among the sub-group of those with a partisan identity (Wagner, Reference Wagner2021), so political engagement determines affective polarization more than elsewhere (on this topic, see also Krupnikov and Ryan, Reference Krupnikov and Ryan2022).
Second, the USA is a clear example of increasing affective polarization, but this trend is not typical of all contexts. For example, Boxell et al. (Reference Boxell, Gentzkow and Shapiro2022) study twelve OECD countries over time, and there were about as many countries that experienced increases as experienced decreases in affective polarization; similar results are reported for a larger number of countries by Garzia et al. (Reference Garzia, Ferreira da Silva and Maye2023) and for Nordic countries by Ryan (Reference Ryan2023). The lack of an overall trend towards affective polarization also means that some macro-level explanations – such as increased internet access or the growth of social media – can only be part of the explanation, if at all (Boxell et al., Reference Boxell, Gentzkow and Shapiro2022). A lack of direct effect of social media use on affective polarization was also not found in individual-level analyses in the Netherlands (Nordbrandt, Reference Nordbrandt2023).
Researching affective polarization in multi-party systems such as in Europe can also help us understand what it is about parties that increases or decreases people’s hostility. For example, the extent to which that party is perceived as a threat (Renström et al., Reference Renström, Bäck and Carroll2021) will determine the extent to which citizens dislike that party’s supporters. This threat is likely to be largely the result of ideological distance: the further away from a citizen’s preferences the plans and policies of an out-party, the more it will be seen as a threat (Kawecki, Reference Kawecki2022; Van Erkel and Turkenburg, Reference Van Erkel and Turkenburg2022; Algara and Zur, Reference Algara and Zur2023). This applies to distance both on economic policy preferences, but also and increasingly on cultural topics (Gidron et al., Reference Gidron, Adams and Horne2023). Riera and Madariaga (Reference Riera and Madariaga2023) find that the link between ideological and affective polarization is stronger in countries that are ideologically polarized and have older democracies, parliamentary systems and smaller party systems. To a certain extent, a strong correlation between ideological and affective polarization is problematic, since it then becomes important to show what studying affective polarization adds to what we know based on studying ideological polarization.
Research on why people dislike some parties more than others is related to work on negative partisanship and out-party hostility (Bankert, Reference Bankert2021). Negative partisanship is largely a synonym of negative out-party affect, but can also encompass a negative party identity, where individuals define themselves by what they are not (Bankert, Reference Bankert2021). The concept therefore turns the spotlight away from polarization per se, that is, the spread of affect. Moreover, negative partisanship does not require the simultaneous presence of positive feelings or identification towards an in-party. Instead, this literature focuses on the strongly disliked parties. Negative partisanship, the presence of which is influenced by the institutional context (Anderson et al., Reference Anderson, McGregor and Stephenson2022), reduces the probability of voting for a party over and above the influence of other factors (Mayer, Reference Mayer2017). Negative partisanship is correlated with lower satisfaction with democracy, especially if negative partisanship is towards major parties (Ridge, Reference Ridge2022). However, a key finding is that negative partisanship in Europe is often towards the radical right (Meléndez and Kaltwasser, Reference Meléndez and Kaltwasser2021; Bjånesøy, Reference Bjånesøy2023), and in this case, negative partisanship can however be a bolster for democracy (Meléndez and Kaltwasser, Reference Meléndez and Kaltwasser2021). While these findings are important, research on negative partisanship in multi-party systems should, where possible, show when and where it is based on a social identity, and the terms out-party hostility or negative out-party affect may be more useful and accurate when such a social identity is not present (on measurement, see Rosema and Mayer (Reference Rosema, Mayer, Oscarsson and Holmberg2020) and Bankert (Reference Bankert2021)).
Moreover, one advantage of comparative research on affective polarization is the possibility of examining how it relates to macro-level factors. Studies on affective polarization in Europe thus highlight how institutional arrangements influence affective polarization: for instance, majoritarian political systems lead to contexts less conducive to elite cooperation, perhaps creating a better foundation for inter-party hostility (Gidron et al., Reference Gidron, Adams and Horne2020). Institutional contexts shape elite behaviour, and how elites interact is clearly central for understanding affective polarization (McCoy and Somer, Reference McCoy and Somer2019; Bassan-Nygate andWeiss, Reference Bassan-Nygate and Weiss2022; Bäck et al., Reference Bäck, Carroll, Renström and Ryan2023). Thus, there is observational and experimental evidence that coalitions between parties can reduce the extent to which partisans dislike each other: supporters of parties that govern together feel more warmly towards each other, even controlling for ideological closeness (Horne et al., Reference Horne, Adams and Gidron2023). Interestingly, this interpartisan warmth persists even after the coalitions end: both current and past co-governance positively influence inter-party affect. Recent experimental work provides additional support for this hypothesis (Praprotnik and Wagner, Reference Praprotnik and Wagner2023). More broadly, cross-national evidence also shows that the parliamentary presence of women in a political party tends to decrease levels of partisan hostility towards that party (Adams et al., Reference Adams, Bracken, Gidron, Horne, O’Brien and Senk2023).
One final result from studying affective polarization in Europe is that in many countries patterns of affect among citizens tend to fall into larger groups. In the USA, many conflicts tend towards a common binary, opposing (liberal, highly educated, urban) Democrats and (conservative, less educated, rural) Republicans (Mason, Reference Mason2016). While partisan divides in Europe are more diverse, studying patterns of affect reveals the existence of broader political camps: parties form ‘affective blocs’ (Kekkonen and Ylä-Anttila, Reference Kekkonen and Ylä-Anttila2021; Bantel, Reference Bantel2023). For example, parties on the left that often form coalitions together may not be seen as rivals but as a larger in-group (‘the left’). Hence, even though there are usually more parties in European countries than in the USA, the key patterns of affective polarization nevertheless reduce to a small number of distinct camps.
What studying affective polarization can tell us about political conflict in Europe
At first glance, the affective polarization framework might seem to provide researchers interested in political conflict in Europe with little insight. After all, Europe is a continent that is largely characterized by declining levels of partisanship (Heath, Reference Heath, Fisher, Fieldhouse, Franklin, Gibson, Cantijoch and Wlezien2017). In many countries, such as some of those in Central and Eastern Europe, positive partisanship is comparatively low (Rose and Mishler, Reference Rose and Mishler1998). Thinking about the mutual dislike between partisan supporters could be interpreted as harking back to bygone eras. Perhaps affective polarization was more applicable to societies in the 50s and 60s, when countries such Austria and the Netherlands were characterized by extremely strong partisan organizations (Lorwin, Reference Lorwin1971). Or perhaps affective polarization applied most to societies in the inter-war period – such as Germany and Spain – that were riven by ideologically driven civil conflict (Bermeo, Reference Bermeo2003). Moreover, critics may argue that in Europe the strongest levels of affect are reserved for other kinds of distinctions: between left and right (Bantel, Reference Bantel2023), between pro- and anti-Brexit (Hobolt et al., Reference Hobolt, Leeper and Tilley2021), or between those who support or oppose Catalan independence (Balcells and Kuo, Reference Balcells and Kuo2023), for instance. These might be the labels citizens actually use when thinking about others’ political affiliations, and they might also be the groups that create stronger stereotypes. However, the fact that affective polarization was likely higher in the past and may be useful to explain other types of distinctions does not mean that partisan affective polarization does not have much to teach us about political conflict in today’s democracies.
Most importantly, research on affective polarization in Europe shows that partisan labels still mean a lot to voters: when people see or hear that someone supports or votes for a party, this creates strong in-group bias (Westwood et al., Reference Westwood, Iyengar, Walgrave, Leonisio, Miller and Strijbis2018; Helbling and Jungkunz, Reference Helbling and Jungkunz2020), at least as strong as ethnic group favouritism. Partisanship and people’s political stances do not leave people cold.
A party family that elicits particularly strong reactions are the radical right (Helbling and Jungkunz, Reference Helbling and Jungkunz2020; Reiljan and Ryan, Reference Reiljan and Ryan2021; Harteveld et al., Reference Harteveld, Mendoza and Rooduijn2022; Gidron et al., Reference Gidron, Adams and Horne2023). Gidron, Adams and Horne (Reference Gidron, Adams and Horne2023) demonstrate that radical-right dislike is even higher than their ideological distance and lack of government participation would predict. The radical left is not disliked to a similar degree, but rather largely forms part of the left camp more generally (Bantel, Reference Bantel2023). Harteveld et al. (Reference Harteveld, Mendoza and Rooduijn2022) argue that the uniquely strong dislike towards the radical right stems from their combination of nativism with populism, both of which divide the population in binary groups: either natives versus non-natives or the ‘elite’ versus the ‘people’ (see also Reiljan and Ryan, Reference Reiljan and Ryan2021). This Manichean thinking is conducive to outpartisan hostility, and not just among radical-right supporters themselves. Social norms and social stigma may play an additional role, as radical right parties are clearly labelled as beyond the pale by mainstream actors in many systems (Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and van der Brug2019). Public tolerance of far-right parties and their supporters thus varies, likely based on the extent to which these parties are clear outsiders excluded from standard political competition or instead regularly included in governing arrangements (Bjånesøy et al., Reference Bjånesøy, Ivarsflaten and Berntzen2023). However, more work needs to be done to explore why the radical right is uniquely disliked. One potential avenue to explore are social norms: perhaps openly expressing dislike towards radical right supporters is seen as more acceptable than dislike towards other supporters, and perhaps such expressions of dislike are even socially desirable in certain environments.
Finally, affective polarization likely has systemic consequences, and comparative research highlights these patterns. One fear is that affective polarization may damage democracy, with McCoy and Somer (Reference McCoy and Somer2019) and McCoy et al. (Reference McCoy, Rahman and Somer2018) even including affective polarization as part of a broader set of ‘pernicious’ developments that endanger democratic quality and stability. Affective polarization has been found to correlate with democratic backsliding (Orhan, Reference Orhan2022), possibly because partisan identities and out-party hostility mean voters are less likely to sanction undemocratic behaviour by politicians they support, partly because of the increased perceived stakes of electoral outcomes (Ward and Tavits, Reference Ward and Tavits2019). Affective polarization also appears to be associated with low social and institutional trust, for example in Spain (Torcal and Thomson, Reference Torcal and Thomson2023) and Sweden (Reiljan and Ryan, Reference Reiljan and Ryan2021). Affective polarization also exacerbates winner–loser gaps (Janssen, Reference Janssen2023) and the tendency to support norm-breaking escalation of political conflict (Berntzen et al., Reference Berntzen, Kelsall and Harteveld2023). However, the research on the link between affective polarization and democratic stability is still in its infancy. Positive effects – such as on political engagement (Ward and Tavits, Reference Ward and Tavits2019; Harteveld and Wagner, Reference Harteveld and Wagner2023) and on exclusion of anti-democratic actors (Meléndez and Kaltwasser, Reference Meléndez and Kaltwasser2021; Reiljan and Ryan, Reference Reiljan and Ryan2021) – still need to be explored more fully.
Four challenges
In this review, I set out four challenges that research on affective polarization in Europe – and in comparative, multi-party settings more generally – should tackle. These encompass questions of measurement, conceptualization, relevance and implications. While many of these points also apply to a certain extent to research on the USA, they are particularly relevant to the European context, where more work needs to be done to measure affective polarization well, prove its relevance, embed it in broader theorizing and discuss its implications. In presenting these challenges I highlight shortcomings rather than solutions.
We need appropriate measures
First, research examining affective polarization in multi-party settings should focus more on testing existing measures and developing appropriate new ones (Gidron et al., Reference Gidron, Sheffer and Mor2022). Thus, the strong use of like–dislike scales in existing research is concerning for two reasons. First, these scales have been in widespread use long before the advent of research on affective polarization. In this research, they have been used as summary evaluations of political parties that contain an ideological and a valence component (Shikano and Nyhuis, Reference Shikano and Nyhuis2019). It is not satisfying to have one single survey measure that has to capture both affective polarization and general party evaluations. While research shows that these scales do correlate with partisan affective polarization (Harteveld, Reference Harteveld2021a) and capture outpartisan hostility (Gidron et al., Reference Gidron, Sheffer and Mor2022), there is also a substantive core to the difference between party and partisan like-dislike measures; for instance, social sorting increases the extent to which like–dislike scores towards parties and partisans correlate (Comellas Bonsfills, Reference Comellas Bonsfills2022).
A second concern is that this measure insufficiently captures what researchers are actually interested in when studying affective polarization. If we are interested in emotional reactions to partisan labels, we should measure this more explicitly (Balinhas, Reference Balinhas2023); if we are instead interested in identities, there are measures for this as well (Huddy et al., Reference Huddy, Mason and Aarøe2015; Bankert et al., Reference Bankert, Huddy and Rosema2017). But arguably, researchers value the concept of affective polarization because it relates to political behaviour and to democratic attitudes. Recently, research by Broockman et al. (Reference Broockman, Kalla and Westwood2023) has shown that experimentally increasing affective polarization (in the form of feeling thermometers) has no effect on democratic attitudes (for a broader perspective on this, see Brauer, Reference Brauer2023). It is quite reasonable to dislike those who do not share our views, especially if these views are dear to us, but treating these people differently in real-world contexts is another matter entirely, and even here the normative questions are tricky. If we justify our interest in affective polarization by its consequences, we may need to focus on those aspects of affective polarization that indeed have the downstream consequences we claim for it.
Another limitation of existing research is that there are two components of affective polarization, positive in-party affect and negative out-party affect, and these should not be conflated (Bankert et al., Reference Bankert, Huddy and Rosema2017). We now have a set of different measures of affective polarization in multi-party systems, and these do a good job at capturing the gap between these two components, with varying approaches and assumptions. However, often researchers are mainly interested in out-party dislike, that is, the extent to which supporters of one party are viewed negatively and treated as an out-group. Because most research uses the term ‘affective polarization’, researchers reach for the polarization measure even when simple out-party dislike could be more appropriate. When designing their analysis, researchers should be careful in thinking about whether they are truly interested in the gap between in-party and out-party affect or whether their theoretical concerns focus only on out-party dislike. In this case, the advantage is that the measures to be used are even simpler, not requiring decisions on the number of in-parties or whether vote share weighting is necessary.
We need a general theory
Second, the field of affective polarization research is in need of a more general theory of the role of group identities for politics and society. Partisan identities are not the only politically relevant identities, and a substantial literature has been developed that examines these in detail (on this point, see Röllicke, Reference Röllicke2023). Most notably, this includes the work of Hobolt et al. (Reference Hobolt, Leeper and Tilley2021) on Brexit identities, but also encompasses political identities based on COVID, for example, in the form of support and opposition to containment measures and vaccination (Bor et al., Reference Bor, Jørgensen and Petersen2023; Henkel et al., Reference Henkel, Sprengholz, Korn, Betsch and Böhm2023) and based on Catalan pro- and anti-independence stances (Hierro and Gallego, Reference Hierro and Gallego2018; Balcells and Kuo, Reference Balcells and Kuo2023) or European identities (Hahm et al., Reference Hahm, Hilpert and König2023). Relevant political identities could also encompass democracy and the party system itself, with social identities forming around opposition or support to political elites or democratic institutions (Moreno, Reference Moreno2019; Meléndez, Reference Meléndez2022; Gessler and Wunsch, Reference Gessler and Wunsch2023; Schedler, Reference Schedler2023). Populism and anti-populism itself can also provide the basis for social divides (Moffitt, Reference Moffitt2018). Such political identities may be particularly likely to develop where there is a strong political tradition of anti-establishment rhetoric and ideology or where there is (a threat of) democratic backsliding.
What this research so far lacks is a general account of how such in- and out-group identities develop and how they lose societal relevance. Existing evidence shows that affective polarization is heightened around elections, partly due to the increase in partisan attachment (Hernández et al., Reference Hernández, Anduiza and Rico2021; Rodríguez et al., Reference Rodríguez, Santamara and Miller2022). Partisan and other political identities can become defining personal characteristics during times of intense conflict, such as the Catalan referendum, the post-Brexit debates and the Covid crisis. Yet, the importance of political identities can also decline, as has clearly happened in the case of Brexit and Covid, and as may occur around national holidays (Levendusky, Reference Levendusky2018). The process by which certain political identities – partisan or policy-based – become dominant and then disappear from relevance deserves further study. It may be that times when these identities are at the forefront of people’s minds are the exception and not the rule. Indeed, most people prefer not to talk about politics and prefer to avoid those who do (Krupnikov and Ryan, Reference Krupnikov and Ryan2022).
Research on affective polarization also needs to engage more with the established literature on cleavages, identities and group conflict in Europe, which has recently experienced a revival of academic interest (Zollinger, Reference Zollinger2024). Related work by Bornschier et al. (Reference Bornschier, Häusermann, Zollinger and Colombo2021) highlights that objective characteristics relate to social identities, often culturally connoted, and that these identities can then be politicized (see also Marks et al., Reference Marks, Attewell, Hooghe, Rovny and Steenbergen2023). Bradley and Chauchard (Reference Bradley and Chauchard2022) show that countries with deeper ethnic divisions also have higher levels of affective polarization. Harteveld (Reference Harteveld2021 b) shows, using both CSES and Dutch panel data, that social sorting – that is, the alignment of political and non-political divisions – increases affective polarization, thus providing a link between affective polarization and social cleavages. The kind of affective polarization between partisans so often studied is thus likely to have its roots in other identities, and these interconnections deserve further study. This work also ties in well with a recent focus on the role of groups in party discourse (Evans and Tilley, Reference Evans and Tilley2017; Thau, Reference Thau2021; Huber, Reference Huber2022; Dolinsky, Reference Dolinsky2023).
These strands of research are highly related, but their interconnections would benefit from being made more explicit and carefully theorized. Work on affective polarization provides the measurement tools and socio-psychological theoretical background to better understand how and when group politics matters, while work on group politics provides the sociological underpinnings for the group attitudes studied by those working on affective polarization. For example, partisan and other political identities may have strong ties to social groups: when people think of those who support certain ideas, they may have stereotypical images of these supporters in their heads. These stereotypes and their implications deserve increased study.
We need evidence of relevance
A key challenge to work on affective polarization is to establish when partisan affect matters in real-world settings. While individuals are able to provide evaluations of groups and react to such descriptions, this does not mean that these characteristics have any relevance in the real world, in everyday interactions. Thus, on a survey, respondents are likely to be able to state whether they hold positive or negative feelings about many groups, including partisans. Similarly, in a vignette experiment respondents may react positively or negatively to partisan identities. For example, Stoetzer et al. (Reference Stoetzer, Munzert, Lowe, Çal, Gohdes, Helbling, Maxwell and Traunmueller2023) show that partisan labels determine the willingness to allocate intensive medical care to individuals. Finally, even tasks with behavioural outcomes such as trust or dictator games will show that people take decisions based on ascribed partisan identities. At the surface, such evidence seems to point to a strong – and potentially worrying – pattern, in that they demonstrate extensive in-group–out-group thinking and the potential for widespread prejudice and discrimination.
However, these findings only show that people react to partisan labels when explicitly provided. Moreover, we know far less about the extent to which people would request that information when encountering new people and thus about how relevant and salient these identities are (Orr and Huber, Reference Orr and Huber2020). Research by Krupnikov and Ryan (Reference Krupnikov and Ryan2022) on the USA also highlights that survey respondents believe that people described as partisan are thought to discuss and talk about politics a lot – and, not surprisingly, this is not something viewed positively by those who do not study and research politics for a living.
Moreover, partisan identities also signal many other characteristics, most obviously political ideology and policy preferences. Interestingly, Hrbková et al. (Reference Hrbková, Voda and Havlk2023) show that immigration attitudes and partisanship both have similar effects on inter-group sympathy in Czechia, so perhaps the policies underlying the partisan label matter just as much as or even more than the group identities. Hence, what respondents react to may not be the identities people have, but rather their ideological stances and the policies they support. One aspect that existing work needs to work much harder at is disentangling (to the extent this is possible) the preferences and attitudes of others from their identities and social group allegiances.
Hence, findings that show affective polarization to be similar across countries need to be interpreted with caution, as the societal relevance of partisan identities may vary widely across contexts. In the USA, identifying oneself as a (loyal) Republican or Democrat might be common and describing people based on this characteristic may occur a lot. However, it is likely that this is much less frequent in other countries, particularly given the declining levels of partisanship found in most European countries (Heath, Reference Heath, Fisher, Fieldhouse, Franklin, Gibson, Cantijoch and Wlezien2017).
As a result, we are left with the possibility that existing results only provide the illusion of partisan affective polarization and that the respondents taking part in our surveys use partisanship as a heuristic for other, more relevant characteristics and would rarely think about partisan identities outside the artificial context we create for them. Future work on partisan affective polarization needs to work harder to show when and why partisan identities matter for people and whether these identities go beyond the policy preferences associated with party support.
We need more normative reflection
A final challenge relates to the normative status of affective polarization. Research on this topic is often justified in terms of the purported negative implications for liberal democracy (McCoy et al., Reference McCoy, Rahman and Somer2018). For instance, recent work shows that the winner–loser gap in political support is greater among affectively polarized citizens (Janssen, Reference Janssen2023). However, there are reasons to challenge this view. For one, the debate is reminiscent of claims that political behaviour should ideally be as devoid of emotional reactions and influences as possible. Yet, as Damasio (Reference Damasio1994) already highlighted, conceiving of decision-making in the absence of emotions is impossible, as emotions are inherent to our thinking and even necessary for us to be able to take decisions. All evidence from social psychology suggests a general human tendency to divide individuals into groups. Of course, these group divisions need not all be conflictual and need not have political implications. Nevertheless, it is hard to conceive of political debates without such group divides: after all, every policy debate separates us into supporters and opponents. If, as Schattschneider (Reference Schattschneider1960) argues, democracy is unthinkable save in terms of political parties, then democracy is also unthinkable save in terms of affective polarization.
Moreover, affective polarization can also have positive consequences. Research on emotions notes that these motivate action (Valentino et al., Reference Valentino, Brader, Groenendyk, Gregorowicz and Hutchings2011). Similarly, affective polarization can also motivate political engagement, and indeed there are close links between affective polarization and key emotions such as anger (Webster, Reference Webster2020). Thus, Harteveld and Wagner (Reference Harteveld and Wagner2023) find that affective polarization increases turnout, with Ellger (Reference Ellger2023) showing that it is mainly negative affect that fosters mobilization (see also Serani, Reference Serani2022; Ahn and Mutz, Reference Ahn and Mutz2023).
Another, more debatable positive aspect of affective polarization concerns the legitimacy of its targets. Many feel uncomfortable with the claim the negative affect towards supporters of specific parties is inherently normatively troubling. The argument is that these supporters have chosen to be loyal to this party, so they have freely decided to agree with the aims and values of that party. In other words, partisanship is an acquired identity. Furthermore, it is arguably legitimate to feel negative affect towards people who hold policy stances that we disagree with. So, negative affect towards partisans does not have the same normative status as, say, negative affect based on race, gender or ethnicity. This argument is even stronger when the parties defend values that are inherently threatening, either to liberal democracy as a whole or to the liberties and rights of certain groups, for example, immigrants or LGBTQI individuals. For example, negative party identity towards the radical right is associated with strong support for democratic values (Meléndez and Kaltwasser, Reference Meléndez and Kaltwasser2021). Here, it is important to relate affective polarization to the concept of militant democracy: some forms of negative affect are arguably legitimate when it comes to defending core democratic and liberal values (Capoccia, Reference Capoccia2013).
This discussion becomes relevant when thinking about ways to reduce affective polarization. In the USA, research on ways to reduce affective polarization and partisan hostility has progressed a lot, allowing for robust inferences on what works and what does not (Hartman et al., Reference Hartman, Blakey, Womick, Bail, Finkel, Han, Sarrouf, Schroeder, Sheeran, Van Bavel, Willer and Gray2022; Voelkel et al., Reference Voelkel, Chu, Stagnaro, Mernyk, Redekopp, Pink, Druckman, Rand and Willer2023). Nevertheless, there is a need for a lot more research on this in multi-party systems. What is important is that research also focuses on the downstream consequences of affective polarization that should be prevented (Brauer, Reference Brauer2023). It is not clear, for example, that measures to reduce affective polarization improve democratic attitudes (Broockman et al., Reference Broockman, Kalla and Westwood2023; Voelkel et al., Reference Voelkel, Chu, Stagnaro, Mernyk, Redekopp, Pink, Druckman, Rand and Willer2023), and it is perhaps those we should care about more than simple negative affect.
Future work on how to reduce affective polarization should thus discuss when and why affective polarization needs to be minimized in the first place. In addition, research needs to be clearer on what the negative consequences of affective polarization are. Potentially, research should focus on addressing either those consequences directly or examine whether reductions in affective polarization also have knock-on effects on those consequences.
Conclusion
Efforts to study affective polarization and partisan hostility in Europe have already paid off. We now know a lot more about levels and trends in affective polarization, about differences between ways of measuring it, and about its correlates at the individual and country level.
Yet, there is still a lot of work to be done. Particularly, three important questions remain. First, how do partisan identities relate to other social identities and cleavages? Second, when and where is partisanship a relevant characteristic in social and political interactions? And third, how can – and why should – we reduce partisan hostility and its potential negative consequences? As I have shown, work on these questions is on-going and will likely provide important, relevant results that will tell us more about both affective polarization and about European politics.
Open access
