Appraising Measurements of Affective Polarisation in Multiparty Systems: Comparative Insights from the Low Countries

• 1 Introduction

  Over the past years, affective polarisation has gained widespread attention and discussion among academics and the general public (Iyengar et al., 2012). Referring to the extent to which citizens feel positively towards co-partisans and negatively towards out-partisans, it is now understood to eclipse ideological considerations (Iyengar et al., 2019). Several worrisome consequences have been associated with affective polarisation, at the political level, increasing political dysfunction and gridlock (Reiljan, 2020) and decreasing satisfaction with democracy (Ridge, 2020), and, at the interpersonal level, inciting heightened tensions and incivility among citizens (Lelkes & Westwood, 2017; Martherus et al., 2019; Westwood et al., 2018).
  European research has more recently turned towards studying the phenomenon of affective polarisation, leaving several domains understudied. One particular aspect is the lack of consensus on measurement approaches. Various measurements are currently used interchangeably, without sufficient explanation why certain measures are preferred or chosen over others. Consequently, researchers may inadvertently tap into different aspects of affective polarisation, which severely hinders our understanding of this phenomenon. Although research has proposed ways of operationalising affective polarisation in multiparty systems (Reiljan, 2020; Wagner, 2021), and confirmed its validity in different national contexts (Russo et al., 2023; Tichelbaecker et al., 2023), much remains unexplored regarding the testing of different operationalisations.
  To address this gap, this study aims to conduct a concurrent validity test by juxtaposing multiple measurements of affective polarisation, building on previous conceptual (Röllicke, 2023) and empirical (Areal & Harteveld, 2024; Harteveld, 2021; Renström et al., 2022; Tichelbaecker et al., 2023) work. Two operationalisations central to the polarisation literature are investigated: like-dislike scores (the affect expressed to specific parties and voters) and social distance (the hesitation to interact with out-partisans). It also includes social avoidance (the tendency for individuals to avoid out-partisans altogether) and out-group dislike (the negative affect expressed towards out-partisans and out-parties), neither of which has so far been examined in conjunction with other measures. This selection does not solely refer to an evaluation (like/dislike); it refers also to intended behaviour (distance/avoidance). In line with previous research, measurements of both the vertical (parties) and the horizontal (voters) dimensions are compared (Harteveld, 2021). As a final validity check, this study explores the different dimensions they capture by correlating them with key drivers and outcomes associated with affective polarisation. It draws on data collected in Belgium and the Netherlands (N = 2,174), two ideal cases to study measurements in a context highly divergent from the US, as they are multiparty systems with high, albeit varying, levels of fractionalisation.
  The results reveal that while the measurements share common factors, their drivers are not identical. In a similar vein, the impact on alleged consequences of affective polarisation varies strongly. Only out-group dislike and social distance are linked to decreased satisfaction with democracy. Affective polarisation measured as the difference between in- and out-group affect does not correlate with democratic satisfaction, but it does seem to stimulate voting intentions. As different measures operate differently both as a dependent variable and an independent variable, concurrent validity may not be as high as some would tend to believe. In sum, this study offers valuable insights with broader relevance to the study of affective polarisation in European multiparty systems. If researchers are mainly interested in studying the negative consequences of affective polarisation, out-group dislike and social distance seem to be more appropriate. Social avoidance, on the other hand, displays particularly low concurrent validity, suggesting that it should be considered a separate dimension. Although more research is needed to uncover which measurements manage to best capture affective polarisation, researchers are strongly cautioned against interchangeably using different measurements in light of their low concurrent validity. Instead, ample thought should be put in the selection of the measurement.

• 2 Conceptualisations and Operationalisations of Affective Polarisation

  Since the past decade, affective polarisation has increasingly become a focal point of political behavioural research. Its theoretical foundations originate from Social Identity Theory, which states that individuals use group membership to navigate social reality (Robinson, 1996; Tajfel et al., 1971; Tajfel & Turner, 1979). In particular, individuals view the world through an ‘in-group’ that they consider themselves a part of, and an ‘out-group’, referring to everyone else. Subsequently, people couple positive emotions to their in-group and negative emotions to the out-group (Sherif et al., 1988; Tajfel, 1970). Political groups are no exception (Mason, 2018b). Iyengar and his colleagues indeed show that US citizens are increasingly affectively polarised (2012). Affective polarisation also seems to be rising in several Western European countries, such as Germany and the United Kingdom (Garzia et al., 2023; Knudsen, 2020; Reiljan, 2020). Hence, some researchers have turned their attention towards the measurement of affective polarisation. The following section will lay out numerous difficulties as well as conceptual ambiguities that scholars face when tackling this issue.

  2.1 Measuring Affective Polarisation in Multiparty Contexts

  Multiple measurements were developed in the US with which researchers have studied affective polarisation (for an overview, see Druckman & Levendusky, 2019). The most popular is the like-dislike or feeling thermometer, which asks respondents to rate parties or voters from strongly dislike to strongly like or cold to warm, respectively (Iyengar et al., 2019). However, when trying to adapt these US-developed measures to Europe, two particular challenges arose: (1) operationalising affective polarisation in multiparty systems and (2) testing measurement validity in highly diverging socio-political contexts.
  In the US, scholars simply have to compute the difference between the level of in-group favouritism and out-group animosity to arrive at someone’s level of affective polarisation (Iyengar et al., 2012). Multiparty systems require a more sophisticated method. Multiple approaches have been proposed since, the most common of which is the Weighted Affective Polarisation (WAP) Index, which creates a sophisticated spread of the affective scores for all citizens (Wagner, 2021). This reflects the notion that voters continue to view the party system through two opposing camps, of which they favour one (Bantel, 2023). It has the distinct advantage of weighing affective polarisation according to party size, which better captures a society’s affective polarisation as the size of a party mirrors its importance in the political arena (Wagner, 2021, p. 3), and it provides scores for all citizens rather than for partisans only (Reiljan, 2020, p. 381).
  Researchers may however inadvertently tap into different aspects of affective polarisation when blindly adopting US-developed scales. To test their validity in multiparty contexts, Russo et al. (2023) compared several operationalisations of affective polarisation using a diverse European student sample spanning nine countries and found that they hold a strong cross-cultural applicability. Similarly, Gidron et al. (2022) validate the feeling thermometer as a measure of partisan affect in Israel’s multiparty system. They showcase that there is indeed a strong overlap between the feeling thermometer towards party supporters, social distance measures and discrimination in economic games.

  2.2 Horizontal versus Vertical Affective Polarisation

  In one of the first conceptual works on affective polarisation in multiparty systems, Röllicke (2023) highlights a number of important ambiguities that remain in the literature. One central such ambiguity is the object of dislike. Although commonly understood as the difference between in-group favouritism and out-group animosity displayed towards political parties or their voters, affective polarisation vis-à-vis political parties and voters is not identical. In particular, animosity towards out-parties does not necessarily translate to similar levels of animosity towards out-party voters (Areal & Harteveld, 2024; Harteveld, 2021). Instead, respondents tend to think of party elites when rating parties, with party elites receiving more negative scores than party voters (Druckman & Levendusky, 2019; Knudsen, 2020).
  Recently, scholars have increasingly utilised two terms to make this distinction: the vertical or political dimension versus the horizontal or social dimension. The vertical dimension pertains to the affect displayed by citizens towards parties or party leaders, whereas the horizontal dimension looks at partisans or party supporters among each other, although it can also examine intergroup affect of ideological camps or other political and issue groups (Comellas & Torcal, 2023; Reiljan & Ryan, 2021; Röllicke, 2023). Whether scholars should focus on one or the other depends on what facet of affective polarisation they are interested in (Areal & Harteveld, 2024). Their consequences are also believed to differ. Horizontal affective polarisation (voters) is said to affect social interactions and lead to ideology-based discrimination, whereas vertical affective polarisation (party elites) should mostly impact the political sphere, such as contributing to political gridlock (Peters, 2021, p. 26). Even though the vertical dimension has been criticised in the US, as it conflates general dislike towards politics with dislike towards specific parties (Klar et al., 2018; Krupnikov & Ryan, 2022), comparing the nuts and bolts of these dimensions in multiparty settings has only recently started (e.g. Gidron et al., 2022; Tichelbaecker et al., 2023). One such reason could be that the two dimensions are more alike in two-party systems where only one partisan out-group exists, resulting in voters more readily extrapolating their affect towards party elites to voters (Areal & Harteveld, 2024). Whether the relationship between these two dimensions is unidirectional, or reciprocal, remains an important gap in the literature.

  2.3 Shallow versus Entrenched Negative Affect

  Despite the prevalence of feeling thermometers and like-dislike scales in the literature, they tend to capture a rather shallow version of affective polarisation (Huddy & Yair, 2021; Kingzette, 2021), reflecting an evaluation rather than true affect or emotion (Verplanken et al., 1998). As a result, strongly disliking an opponent may be multidimensional in itself, ranging from mere dislike to feelings of deep hatred. Measures exist that allow researchers to disentangle them and distinguish between shallow and entrenched types of negative affect. Here, the latter should tap into more extreme forms of out-group bias that are not as easily expressed as shallow negative affect.
  More entrenched forms of horizontal affective polarisation have previously been captured with social distance items (Iyengar, 2022). These ask respondents how they would feel interacting with out-partisans in different social settings, such as their level of comfort if their child would marry someone from a political out-group (Iyengar et al., 2012, 2019; Mason, 2018b). Social distance indeed seems to better capture a more deep-rooted dislike towards the ‘other side’, as it constitutes a more extreme form of ostracisation (Druckman & Levendusky, 2019). This is reflected in scores being lower overall than for the feeling thermometer or like-dislike scales (Tichelbaecker et al., 2023). Moreover, it has the advantage of capturing (intended) behaviour, which does not necessarily arise from pre-existing levels of affect (Clore & Schnall, 2019; Terry & Hogg, 1996). Similarly, social avoidance, or the general tendency of people to avoid others based on certain characteristics such as their political views, taps into more general conflict avoidance and is strongly placed on the horizontal dimension (Huber & Malhotra, 2017). It may also in part capture the behavioural consequences of affective polarisation (Iyengar et al., 2019).
  Two additional examples of measurements which capture more entrenched forms of dislike are traits and discrete emotions. Trait batteries tend to include both positive and negative items (Druckman & Levendusky, 2019; Kelly Garrett et al., 2014; Renström et al., 2021). Little research has however been conducted on the application of the traits battery in multiparty settings. Similarly, measuring discrete emotions is only remarkably rarely done in European literature (Berntzen et al., 2024; Nguyen et al., 2022; Renström et al., 2023) and receives less focus in the US as well (Webster, 2020; Webster & Albertson, 2022). Nevertheless, political psychology has theorised extensively on which mechanisms drive and influence emotions and affect (Marcus et al., 2000; Redlawsk & Mattes, 2022).

  2.4 Affective Polarisation versus Out-Group Dislike

  Another important ambiguity in the affective polarisation literature, as pointed out by Röllicke, is that “negative out-group evaluations can occur for reasons that have nothing to do with an in-group” (2023, p. 7). Although affective polarisation is commonly conceptualised as the difference between one’s favouritism towards the political in-group(s) and animosity towards the political out-group(s) (Bantel, 2023; Iyengar et al., 2012; Wagner, 2021), a review of the literature reveals that many scholars solely study out-group dislike, omitting affective polarisation’s in-group component (e.g. Gidron et al., 2022; Harteveld et al., 2021; Harteveld & Wagner, 2023; Simas et al., 2020). This interest is likely driven by the fact that negativity bias has been described as more pervasive than positivity bias (Iyengar & Krupenkin, 2018, p. 212), which is theoretically and empirically a distinct phenomenon (Bougher, 2017; Brewer, 1999). Both types of biases are likely driven by different factors. For example, political system fragmentation seems more strongly associated with out-party animosity than in-group favouritism (Gidron et al., 2020, pp. 66-67).
  Indeed, out-group dislike is sometimes considered to be more pertinent than in-group attachment, particularly in a multiparty context (Wagner, 2021, p. 7), which has been discussed extensively in the literature on negative partisanship (Bankert, 2020, 2022; Mayer & Russo, 2023). The reason may stem from the fact that in most multiparty systems, the negative affect of centre-left and centre-right individuals towards one another may not be as high as in the US, where it has become a defining characteristic of American politics (Iyengar et al., 2019). European polarisation is mostly driven by negative affect towards and from the radical right (Harteveld et al., 2021). This suggests that affective polarisation in Europe may be less typified by a tug of war between the left and the right, but more as a clash between mainstream party voters and the radical right (Bantel, 2023). Affective polarisation may therefore be better captured in multiparty systems by solely examining affect towards the out-group.
  Currently, most Europe-developed measurements based on feeling thermometers do not separate the in- and out-party components and instead consider them equal in shaping affective polarisation (Wagner, 2021). Iyengar and his colleagues, however, claim that “the precise mix of in- and outgroup sentiment” may differ depending on an “individuals’ prior information and how they update beliefs based on exposure to new information” (2019). There are reasons to believe that there are situations in which only considering out-group dislike has some merit. For example, Wagner’s index shows that, perhaps counter-intuitively, polarisation decreased in the US between 2012 and 2016 (2021), which is driven by the fact that the decrease in in-group favouritism was stronger than the increase in out-group animosity (Iyengar et al., 2019).
  As a consequence of decoupling the in- and out-groups, we do not study ‘polarisation’ as such anymore. This detaches affective polarisation from its theoretical foundations in Social Identity Theory (Tajfel & Turner, 1986), which are mostly based on an experiment (Robber’s Cave) in which scholars carefully designed the environment to maximise the chance of creating group attachments and triggering intergroup conflict. However, this may not necessarily reflect daily political settings (for a discussion, see Krupnikov & Ryan, 2022), especially in multiparty contexts where multiple in- and out-groups are present. As mentioned previously, scholars of negative partisanship do not consider in-group favouritism a required precondition of out-group animosity. Particularly in contexts where in-group favouritism or partisan attachment is considered low (Huddy et al., 2018), or when the research question is mostly interested in the causes or consequences of out-group animosity, it may be more appropriate to focus solely on out-group dislike, and even though one deviates from studying polarisation in the strictest sense of the word, one could still categorise it under the umbrella term of affective polarisation.

• 3 Data and Methods

  This study relies on a survey conducted in Belgium and the Netherlands in June 2020 by the market research company Respondi. Using computer-assisted web interviews via an Online Access Panel (N = 2,174; Belgium: N = 1,071; the Netherlands: N = 1,103), data were derived from a nonprobability sample with matched quotas for age (5 categories), gender and NUTS-1 region, and 18- to 69-year-olds were sampled, resulting in a nationally representative sample that focuses on people at working age. Most respondents completed the questionnaire in ±15 minutes.
  Belgium and the Netherlands are two ideal cases to study affective polarisation measurements in multiparty systems, as they have a long history of coalition governments and their political systems are highly fractionalised; that is, the effective number of political parties is (very) high in both systems. This makes them strongly diverge from the US, a most well-examined case. Belgium and the Netherlands also differ from one another in important ways. The Netherlands stands out as one of the least affectively polarised countries in Europe (Harteveld, 2021; Wagner, 2021). Parties in Flanders and the Netherlands also take a very different approach to the radical right. This is important in light of the radical right’s centrality in shaping affective polarisation (Harteveld et al., 2021). Whereas parties have gone so far as to cooperate with the radical right in the Netherlands, Belgium (so far) maintains a strict cordon sanitaire (Mudde, 2002). Lastly, the degree of fractionalisation is much higher in the Netherlands, whereas Belgium contains a strong linguistic divide which leads to separate party systems for each region (Deschouwer, 2012).
  Exactly due to the (highly) fractionalised systems present in Belgium and the Netherlands, party selection for the affective polarisation questions was particularly tricky. For Flanders and Wallonia, the survey included all parties with seats in federal parliament (7 and 6, respectively). Wallonia lacks a strong radical right party, whereas Flanders’ radical left party is considerably smaller than Wallonia’s. As the makeup of their party systems thus differs considerably, the Belgian data will be split according to the party system of respondents, resulting in two subsets in the analysis: Flanders (N = 615) and Wallonia (N = 448). As 17 parties were seated in the Dutch national parliament at the time of data collection, different selection criteria needed to be considered. Cognitive strain on respondents would have been enormous, resulting in high drop-out rates, non-response and satisficing. The number of parties was whittled down to the 10 biggest parties, each of which has at least 5 seats and 3% of the votes (VVD as the largest, ChristenUnie as the smallest). This also includes all coalition parties. Although some (very) small parties are excluded that some people may feel particularly strongly attracted towards, their effect on the eventual like-dislike score would be small regardless, as these are weighted by vote share. The parties that are at the ideological extremes and attract the most negative affect (PVV, FvD and SP) are present (Harteveld et al., 2021). Although conceptual validity would increase slightly when including all 17 parties, the reliability of the scale would strongly decrease due to the expected satisficing and non-response. The selection of 10 parties is therefore preferred.
  This study will follow the approach by Druckman and Levendusky (2019) and Russo et al. (2023) by incorporating a large number of the items. The main measurements examined here are the like-dislike scale towards parties and voters, social distance towards a close friend and romantic partner, and social avoidance (response scales: 1-7). The first captures the vertical dimension, whereas the horizontal dimension is measured by the four other measures (Areal & Harteveld, 2024; Iyengar et al., 2019). The first of these horizontal measures refers merely to an evaluation, whereas the last three tap into more entrenched forms of polarisation, namely its (intended) behavioural tendencies (Röllicke, 2023). Social distance was measured towards voters of the respondent’s three ‘least-liked’ parties. Social avoidance was measured by asking respondents to what extent they tend to avoid people based on their political views, similar to items used in previous research (Hetherington & Rudolph, 2015; Lelkes, 2016; McCoy & Somer, 2019).
  The study is also interested in the concurrent validity of out-group dislike. As citizens tend to hold multiple in-group and out-group identities in multiparty systems (Bantel, 2023), one cannot simply use the (dis)favourability rating of the out-party (Simas et al., 2020). To tackle this, this study leverages a novel question which asks respondents to rank all parties according to their favourability, so the analysis can link this question to the like-dislike scale and examine whether patterns emerge depending on negative affect towards one’s first, second or third out-party.
  In addition, the analysis utilises exploratory factor analysis (EFA) to assess whether these different measurements tap into one or multiple latent constructs. Subsequently, a series of linear and logistic regression analyses examine a number of alleged key drivers and consequences of affective polarisation. The said analyses report findings for each measurement for the combined dataset using party-system fixed effects and for each party system separately.
  For the key drivers of affective polarisation, widely believed in the literature to increase affective polarisation, the analysis includes political interest (Banda & Cluverius, 2018; Krupnikov & Ryan, 2022); ideological extremism (Mason, 2018a; Reiljan, 2020; van Erkel & Turkenburg, 2022); positive partisanship, that is, the extent to which an individual identifies with a certain party (Harteveld, 2021; Hobolt et al., 2020; Iyengar & Westwood, 2015; Wagner, 2021); and negative partisanship, that is, the extent to which one is repulsed by their out-party (Bankert, 2020; Huddy et al., 2018; Martherus et al., 2019). Positive partisanship is measured through the party closeness question. Negative partisanship relies on a novel two-item battery developed by Mayer and Russo (2023),¹x Agree-disagree: (1) Because of their worldviews, I could never vote for this party. (2) It is important to me that I am not one of those people who vote for this party. which is asked for the most disliked party only.
  Research on the consequences of affective polarisation is less unanimous. The analysis will test the following often-considered outcomes: satisfaction with democracy (Ridge, 2020, 2021), social trust (Hye & Lee, 2022; Torcal & Thomson, 2023), and voting participation (Harteveld & Wagner, 2023). Voting participation is captured through intention to vote if elections were held tomorrow.²x Due to mandatory voting in Belgium, it asked whether respondents would vote if elections were not mandatory. All analyses control for age, gender and education attainment.
  It is important to note that the regression analyses are purely exploratory in nature. There is a potential for reversed causality between most, if not all, affective polarisation measurements and their alleged drivers and outcomes. The analysis will therefore not seek to make causal claims.

• 4 Results

  4.1 Descriptive Statistics

  As we know from previous research, out-group dislike decreases with increasing ideological similarities, and the radical right is uniquely disliked, both in Belgium (van Erkel & Turkenburg, 2022) and the Netherlands (Harteveld, 2021). Before moving to a full comparison of all the measures, this section zooms in on out-group affect. Figure 1 displays the different levels of affective dislike and social distance towards respondents’ three most disliked parties. Negative affect decreases when moving away from the least-liked party, suggesting that the rank-order question functioned as intended.³x According to paired t-tests, these differences are significant (p < 0.001). The degree of negative affect is comparable across the three different party systems, except for social distance, as Wallonia scores consistently lower. Moreover, only in the Netherlands and Flanders, the question on social distance towards a romantic partner scores higher than towards a close friend. Overall, scores within each measurement do not differ substantially, which is confirmed by a correlation analysis (r > 0.84; see Appendix A). Further analysis will combine the separate scores for each party into one for each measurement.
  Averages for each measurements are displayed in Figure 2. For comparability’s sake, measures were rescaled to 0-1. In line with the literature, both the spread measure (referred to as WAP, short for Weighted Affective Polarisation Index) and dislike towards parties are slightly higher than for voters (diff. = 0.07-0.08, p < 0.001). As can be expected, the WAP measures are lower than the dislike measures, as they also include the in-group component. Similar patterns are observed as in Figure 1. In addition, respondents tend to exhibit relatively little social avoidance, suggesting that it serves as a more conservative estimate of negative affect.⁴x According to paired t-tests, all pairs of measurements are significantly different (p < 0.01).

  Out-Group Dislike
  
  Affective Polarisation Measurements
  

  Correlation analysis of the main measurements (see Figure 3) shows that most correlations are significant (p < 0.001). The WAP measures towards parties and voters correlate strongly (r = 0.68), as do the out-group dislike measures (r = 0.72). The two social distance measures are also strongly related (r = 0.78, p < 0.001). The connection between the WAP and dislike measures is much weaker (r = 0.10-0.19). Social avoidance stands out, only moderately correlating with social distance towards a close friend (r = 0.22). To further examine the relation between these different measurements, this section now turns to factor analyses.

  Correlation Matrix
  

  4.2 Factor Analysis

  Following Russo et al.’s approach (2023), only scores towards the most disliked party are considered, holding the partisan group constant. This prevents the choice of parties from influencing the results, instead solely focusing on the measurement and object of dislike. Social avoidance is also included, as it probes people’s avoidance towards those holding different political views. As shown in Table 1, two factors are retained (Eigenvalue1 = 2.91; Eigenvalue2 = 1.01). Factor 1 captures dislike and social distance, whereas social avoidance constitutes its own factor. When removing social avoidance, all variables load unto one factor and reliability increases to 0.87. This process is repeated for scores towards the second and third most disliked party (see Appendix B). Results are highly similar, with Eigenvalue2 hovering around 1.00, and an improvement in reliability from 0.69-0.76 to 0.82-0.86 when removing social avoidance. When examining party systems separately, Eigenvalue2 only dips below 1.00 in Flanders (0.96). In sum, whether there are two factors depends on the parties and party systems examined, but social avoidance differs consistently from the other measures and its removal leads to substantially higher reliability ratings.

  Table 1 Explanatory Factor Analysis

  Variable	Factor 1	Factor 2	Uniqueness
  Dislike party	0.78	−0.31	0.29
  Dislike voter	0.86	−0.12	0.24
  Social distance friends	0.85	0.13	0.27
  Social distance partner	0.88	−0.01	0.23
  Social avoidance	0.25	0.94	0.05

  4.3 Regression Analysis: Key Drivers

  The standardised results of four key drivers of affective polarisation are presented in Figure 4 (see Appendix C for full regression results). Political interest seems to matter most for the WAP measures. Ideological extremism and negative partisanship return the most robust results, with significant associations between higher levels of ideological extremism and negative partisanship on the one hand, and higher levels of affective polarisation on the other hand (p < 0.05). For positive partisanship, an interesting pattern can be observed. Albeit not entirely unexpected, it seems to be most consequential for the WAP measure towards parties (p < 0.05) and (slightly less) towards voters (p < 0.10). The social distance and avoidance items are not predicted by positive partisanship (p > 0.05). This should not come as a great surprise, as these items lack an in-group component.⁵x Separate results for each party system are presented in Appendix D.

  Key Drivers (Belgium and Netherlands)
  

  4.4 Regression Analysis: Key Outcomes

  The following section focuses on three key outcomes of affective polarisation: satisfaction with democracy, social trust and voting intention (full results in Appendix C). The standardised results for satisfaction with democracy are included in Figure 5. WAP towards parties is significantly associated with higher satisfaction with democracy (p > 0.05), whereas WAP towards voters is insignificant (p > 0.05). Higher dislike towards parties and voters and increased social distance predict lower satisfaction with democracy (p < 0.001). Results for social avoidance are inconclusive (p > 0.05). For social trust, shown in Figure 6, only the two dislike measures lead to significant reductions with very small effect sizes (p < 0.05). In other words, affective polarisation only seems to play a marginal role in shaping social trust in Belgium and the Netherlands. Lastly, the analysis turns towards the association between affective polarisation and voting intention (see Figure 7). As voting intention is a binary variable, logistic regression analyses are computed and log odds are displayed below. Overall, affective polarisation seems to matter in increasing voting intention, in line with findings by Harteveld and Wagner (2023), but results are not conclusive across measurements. The WAP and dislike scores towards parties, but not voters, are associated with an increased likelihood to vote (p < 0.05), in line with the results for social distance (p < 0.05).⁶x Separate results for each party system are presented in Appendix D.

  Satisfaction with Democracy (Belgium and Netherlands)
  
  Social Trust (Belgium and Netherlands)
  
  Voting Intention (Belgium and Netherlands)
  

• 5 Discussion and Conclusion

  As scholars of political science have increasingly turned towards studying affective polarisation, few studies have yet examined how US-developed conceptualisations and operationalisations extend to (European) multiparty contexts. Researchers nonetheless have a wide variety of measurements at their disposal, but the selection of appropriate measurement methods from the available options remains challenging and often arbitrary. This article aimed to explore how these measurements are interconnected, assessing their concurrent validity and exploring their various dimensions and implications for political behaviour and social cohesion. Leveraging cross-country data from Belgium and the Netherlands (N = 2,174), the present study aimed to uncover nuanced insights into the complexity of affective polarisation, both challenging and extending existing frameworks. As a result, it further adds to an increasing body of research that also aims to inform polarisation research as to which measurement they should pick for their data collection and analysis (Areal & Harteveld, 2024; Russo et al., 2023; Tichelbaecker et al., 2023). It aimed to further expand the palate of measurements available to affective polarisation researchers, as well as examining how each of them is related to one another and under which circumstance(s) which measurement is the most appropriate. Several notable results should be highlighted.
  The study combined several goals. First, it assessed the concurrent validity of several affective polarisation measures – like-dislike scales, social distance and social avoidance – in a multiparty system. The findings reveal that while they share commonalities, they also seem to capture unique facets of affective polarisation. Social avoidance in particular stands apart, further confirming that affect does not necessarily translate into (intended) behaviour (Clore & Schnall, 2019; Terry & Hogg, 1996). However, the relation between behavioural and affective polarisation is not yet well understood, warranting future research to examine whether one is causally prior to the other, for example, whether behavioural polarisation requires some level of affective polarisation. The study also showed that the horizontal and vertical dimensions of affective polarisation are related but are not necessarily driven by similar factors or exert analogous influences. This is in line with previous research (Areal & Harteveld, 2024), validating that researchers should not conflate one with the other. Interestingly, the difference between the vertical dimension and the horizontal dimension is not nearly as pronounced as the difference between the horizontal like-dislike and social distance items, suggesting that the type of measurement matters more than the dimension one is interested in. What exactly drives these differences so far remains unclear. Future research should examine under which conditions, for example, out-party dislike and social distance diverge and what consequences such patterns have downstream. So far, scholars are in the dark about what respondents exactly picture when filling out affective polarisation questions. Qualitative interviews could be particularly well-suited to tackle this challenge.
  Second, the study’s focus on Belgium and the Netherlands, with their similar, yet distinct, political landscapes, served to offer insights into the manifestation of affective polarisation in diverse multiparty systems. Results often differed between party systems. This became particularly evident when focusing on a number of key drivers of affective polarisation. Ideological extremism was the most robust predictor of affective polarisation, regardless of its measurement of context. Positive partisanship only predicted the weighted spread scores, whereas negative partisanship only mattered in Flanders and the Netherlands. The absence of a radical right party in Wallonia might explain this difference, underscoring the diverseness of the three cases. While the results may more broadly apply to other European countries, this study emphasises that the nuances observed in each context are important and may influence results depending on which measurement one uses.
  Third, this article shed light on the relationship between affective polarisation and key political behaviours. The findings indicate varying, and often inconclusive, impacts of affective polarisation on satisfaction with democracy, social trust and voting intentions. Nonetheless, higher levels of out-group dislike and social distance are robustly linked to decreased satisfaction with democracy, signalling the potential erosion of democratic health due to intense partisan animosity (Ridge, 2020). When considering both in- and out-group components of affective polarisation, the analysis found no negative effect. Interestingly, affective polarisation as such even appears to stimulate voting intentions, suggesting a mobilising effect despite its negative connotations (Harteveld & Wagner, 2023).
  This study comes with several limitations. The sample only encompassed three party systems in two countries. Although the varying measures of affective polarisation seemed to operate differently across party systems, statistical power issues prevent the analysis from concluding whether and how these might be conceptually driven. Further research is now needed to examine which contextual factors explain these differences. Moreover, expanding the country selection by including less fractionalised party systems (e.g. Germany) or countries with clear ideological blocks (e.g. Denmark and Sweden) would lead to stronger generalisability. The number of examined causes could also be enlarged to strengthen the concurrent validity test, such as propensity to vote. Lastly, several measures used in affective polarisation research were not examined. Two particular examples are emotions and traits, which strike at a more entrenched form of polarisation, but are quite rarely used. For a concept called affective polarisation, the relative lack of research focusing on the underlying emotions of this affect is surprising. Future research should test their concurrent validity with the other measures examined in this study.
  To conclude, this study contributes to a deeper, more nuanced understanding of affective polarisation in multiparty systems. The generally low levels of concurrent validity highlight the need for a multifaceted approach to comprehensively grasp the phenomenon and for future research to focus on understanding what respondents think when filling out their responses to polarisation questions. The study also confirmed Russo et al.’s finding (2023) that measures in a multiparty context strongly depend on which parties are surveyed and whether researchers only consider the out-group or both in-group and out-group. Although more research is required, this study puts forth the following three pieces of advice: (1) if researchers are interested in studying the negative consequences associated with political intergroup conflict in multiparty systems, they should focus on out-group dislike and social distance. If they are interested in more entrenched forms of affective polarisation, social distance seems the more appropriate measure, keeping in mind that it may be strongly affected by the presence or absence of a radical right party. (2) As social avoidance displayed particularly low concurrent validity, it is better viewed as a distinct measure which captures behavioural or social rather than affective polarisation. Measuring affective polarisation exclusively with social avoidance is not advisable. (3) Researchers should think carefully about what they are trying to measure conceptually, letting these considerations guide the choice as to which measurement(s) to include in a study rather than falling into the trap of post hoc cherry-picking. Although considerable overlap exists, confirmed by previous research (e.g. Druckman & Levendusky, 2019; Russo et al., 2023; Tichelbaecker et al., 2023), this study shows that the different measurements of affective polarisation cannot and should not be used interchangeably.
  In sum, as affective polarisation continues to shape political landscapes and dominate political discourse, further measurements research remains imperative, as it allows researchers to explore the evolving dynamics of affective polarisation in order to prevent the deterioration of political systems into destructive ‘us versus them’ confrontations.

References

• Areal, J. & Harteveld, E. (2024). Vertical vs Horizontal Affective Polarization: Disentangling Feelings towards Elites and Voters. Electoral Studies, 90. https://doi.org/10.1016/j.electstud.2024.102814

• Banda, K. K. & Cluverius, J. (2018). Elite Polarization, Party Extremity, and Affective Polarization. Electoral Studies, 56(April), 90-101. https://doi.org/10.1016/j.electstud.2018.09.009.

• Bankert, A. (2020). Negative and Positive Partisanship in the 2016 U.S. Presidential Elections. Political Behavior, 43, 1467-1485. https://doi.org/10.1007/s11109-020-09599-1.

• Bankert, A. (2022). The Personality Origins of Positive and Negative Partisanship. Politics and Governance, 10(4), 299-310. https://doi.org/10.17645/pag.v10i4.5719.

• Bantel, I. (2023). Camps, Not Just Parties. The Dynamic Foundations of Affective Polarization in Multi-Party Systems. Electoral Studies, 83(April), 102614. https://doi.org/10.1016/j.electstud.2023.102614.

• Berntzen, L. E., Kelsall, H. & Harteveld, E. (2024), Consequences of affective polarization: Avoidance, intolerance and support for violence in the United Kingdom and Norway. European Journal of Political Research. https://doi.org/10.1111/1475-6765.12623

• Bougher, L. D. (2017). The Correlates of Discord: Identity, Issue Alignment, and Political Hostility in Polarized AMERICA. Political Behavior, 39(3), 731-762. https://doi.org/10.1007/s11109-016-9377-1.

• Brewer, M. B. (1999). The Psychology of Prejudice: Ingroup Love or Outgroup Hate? Journal of Social Issues, 55(3), 429-444. https://doi.org/10.1111/0022-4537.00126.

• Clore, G. L. & Schnall, S. (2019). The Influence of Affect on Attitude. In D. Albarracin & B. T. Johnston (Eds.), The Handbook of Attitudes (pp. 437-489). Routledge.

• Comellas, J. M. & Torcal, M. (2023). Ideological Identity, Issue-Based Ideology and Bipolar Affective Polarization in Multiparty Systems: The Cases of Argentina, Chile, Italy, Portugal and Spain. Electoral Studies, 83(April), 102615. https://doi.org/10.1016/j.electstud.2023.102615.

• Deschouwer, K. (2012). The Politics of Belgium. Palgrave Macmillan.

• Druckman, J. N. & Levendusky, M. S. (2019). What Do We Measure When We Measure Affective Polarization? Public Opinion Quarterly, 83(1), 114-122. https://doi.org/10.1093/poq/nfz003.

• Garzia, D., Ferreira, F. & Maye, S. (2023). Affective Polarization in Comparative and Longitudinal Perspective Operationalization Strategies. Public Opinion Quarterly, 87(1), —219-231. https://doi-org/10.1093/poq/nfad004.

• Gidron, N., Adams, J. & Horne, W. (2020). American Affective Polarization in Comparative Perspective. Cambridge University Press.

• Gidron, N., Sheffer, L. & Mor, G. (2022). Validating the Feeling Thermometer as a Measure of Partisan Affect in Multi-Party Systems. Electoral Studies, 80(September), 102542. https://doi.org/10.1016/j.electstud.2022.102542.

• Harteveld, E. (2021). Fragmented Foes: Affective Polarization in the Multiparty Context of the Netherlands. Electoral Studies, 71(March), 102332. https://doi.org/10.1016/j.electstud.2021.102332.

• Harteveld, E., Mendoza, P. & Rooduijn, M. (2021). Affective Polarization and the Populist Radical Right: Creating the Hating? Government and Opposition, 1-25. https://doi.org/10.1017/gov.2021.31.

• Harteveld, E. & Wagner, M. (2023). Does Affective Polarisation Increase Turnout? Evidence from Germany, The Netherlands and Spain. West European Politics, 46(4), 732-759. https://doi.org/10.1080/01402382.2022.2087395.

• Hetherington, M. & Rudolph, T. (2015). Why Washington Won’t Work: Polarization, Political Trust, and the Governing Crisis. Chicago Press.

• Hobolt, S. B., Leeper, T. J. & Tilley, J. (2020). Divided by the Vote: Affective Polarization in the Wake of the Brexit Referendum. British Journal of Political Science. https://doi.org/10.1017/S0007123420000125.

• Huber, G. A. & Malhotra, N. (2017). Political Homophily in Social Relationships: Evidence from Online Dating Behavior. Journal of Politics, 79(1), 269-283. https://doi.org/10.1086/687533.

• Huddy, L., Bankert, A. & Davies, C. (2018). Expressive Versus Instrumental Partisanship in Multiparty European Systems. Political Psychology, 39, 173-199. https://doi.org/10.1111/pops.12482.

• Huddy, L. and Yair, O. (2021). Reducing Affective Polarization: Warm Group Relations or Policy Compromise? Political Psychology, 42, 291-309. https://doi-org.kuleuven.e-bronnen.be/10.1111/pops.12699.

• Hye, A. & Lee, Y. (2022). Social Trust in Polarized Times: How Perceptions of Political Polarization Affect Americans’ Trust in Each Other. Political Behavior, 44(3), 1533-1554. https://doi.org/10.1007/s11109-022-09787-1.

• Iyengar, S. (2022). Fear and Loathing in American Politics: A Review of Affective Polarisation. In D. Osborne & C. G. Sibley (Eds.), The Cambridge Handbook of Political Psychology (pp. 399-413). Cambridge University Press.

• Iyengar, S. & Krupenkin, M. (2018). The Strengthening of Partisan Affect. Political Psychology, 39, 201-218. https://doi.org/10.1111/pops.12487.

• Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N. & Westwood, S. J. (2019). The Origins and Consequences of Affective Polarization in the United States. Annual Review of Political Science, 22, 129-146. https://doi.org/10.1146/annurev-polisci-051117-073034.

• Iyengar, S., Sood, G. & Lelkes, Y. (2012). Affect, Not Ideology: A Social Identity Perspective on Polarization. Public Opinion Quarterly, 76(3), 405-431. https://doi.org/10.1093/poq/nfs038.

• Iyengar, S. & Westwood, S. J. (2015). Fear and Loathing across Party Lines: New Evidence on Group Polarization. American Journal of Political Science, 59(3), 690-707. https://doi.org/10.1111/ajps.12152.

• Kelly Garrett, R., Gvirsman, S. D., Johnson, B. K., Tsfati, Y., Neo, R. & Dal, A. (2014). Implications of Pro- and Counterattitudinal Information Exposure for Affective Polarization. Human Communication Research, 40(3), 309-332. https://doi.org/10.1111/hcre.12028.

• Kelsall, H. (2023). Affective Polarization, (Mis)perceived Polarization and Negative Meta-Perceptions: Pre-Analysis Plan. Politicologenetmaal.

• Kingzette, J. (2021). Who Do You Loathe? Feelings toward Politicians vs. Ordinary People in the Opposing Party. Journal of Experimental Political Science, 8(1), 75–84. doi:10.1017/XPS.2020.9.

• Klar, S., Krupnikov, Y. & Ryan, J. B. (2018). Affective Polarization or Partisan Disdain? Untangling a Dislike for the Opposing Party from a Dislike of Partisanship. Public Opinion Quarterly, 82(2), 379-390. https://doi.org/10.1093/poq/nfy014.

• Knudsen, E. (2020). Affective Polarization in Multiparty Systems? Comparing Affective Polarization towards Voters and Parties in Norway and the United States. Scandinavian Political Studies, 44(1), 1-11. https://doi.org/10.1111/1467-9477.12186.

• Krupnikov, Y. & Ryan, J. B. (2022). The Other Divide. Cambridge University Press.

• Lelkes, Y. (2016). The Polls-Review: Mass Polarization: Manifestations and Measurements. Public Opinion Quarterly, 80(S1), 392-410. https://doi.org/10.1093/poq/nfw005.

• Lelkes, Y. & Westwood, S. J. (2017). The Limits of Partisan Discrimination. Journal of Politics, 79(2), 485-501. https://pdfs.semanticscholar.org/7f2b/76af4199042228aae09a51a04bdecdcd9468.pdf?_ga=2.41911074.1682974136.1539702341-877505766.1534072953.

• Marcus, G. E., Neuman, W. R. & MacKuen, M. (2000). Affective Intelligence and Political Judgment. University of Chicago Press.

• Martherus, J. L., Martinez, A. G., Piff, P. K. & Theodoridis, A. G. (2019). Party Animals? Extreme Partisan Polarization and Dehumanization. Political Behavior, 43(2), 517-540. https://doi.org/10.1007/s11109-019-09559-4.

• Mason, L. (2018a). Ideologues without Issues: The Polarizing Consequences of Ideological Identities. Public Opinion Quarterly, 82(S1), 280-301. https://doi.org/10.1093/poq/nfy005.

• Mason, L. (2018b). Uncivil Agreement: How Politics Became Our Identity. The University of Chicago Press.

• Mayer, S. J. & Russo, L. (2023). What One Is Not: A New Scale to Measure Negative Party Identity in Multiparty Systems. Quality and Quantity, 58, 2887-2906. https://doi.org/10.1007/s11135-023-01793-7.

• McCoy, J. & Somer, M. (2019). Toward a Theory of Pernicious Polarization and How It Harms Democracies: Comparative Evidence and Possible Remedies. Annals of the American Academy of Political and Social Science, 681(1), 234-271. https://doi.org/10.1177/0002716218818782.

• Mudde, C. (2002). The Ideology of the Extreme Right. Manchester University Press. www.ssoar.info.

• Nguyen, C. G., Mayer, S. J. & Veit, S. (2022). The Impact of Emotions on Polarization. Anger Polarizes Attitudes towards Vaccine Mandates and Increases Affective Polarization. Research & Politics, 9(3). https://doi.org/10.1177/20531680221116571.

• Peters, U. (2021). How (Many) Descriptive Claims about Political Polarization Exacerbate Polarization. Journal of Social and Political Psychology, 9(1), 24-36. https://doi.org/10.5964/jspp.5543.

• Redlawsk, D. P. & Mattes, K. (2022). Emotions and Politics. In D. Osborne & C. G. Sibley (Eds.), The Cambridge Handbook of Political Psychology. Cambridge University Press.

• Reiljan, A. (2020). ‘Fear and Loathing across Party Lines’ (also) in Europe: Affective Polarisation in European Party Systems. European Journal of Political Research, 59(2), 376-396. https://doi.org/10.1111/1475-6765.12351.

• Reiljan, A. (2023). What Makes It Personal? The Polynomial Relationship between Political and Social Manifestations of Affective Polarization. EPSA Annual Meeting.

• Reiljan, A., Garzia, D., Ferreira Da Silva, F. & Trechsel, A. H. (2023). Patterns of Affective Polarization toward Parties and Leaders across the Democratic World. American Political Science Review, 1-17. https://doi.org/10.1017/S0003055423000485.

• Reiljan, A. & Ryan, A. (2021). Ideological Tripolarization, Partisan Tribalism and Institutional Trust: The Foundations of Affective Polarization in the Swedish Multiparty System. Scandinavian Political Studies, 44(2), 195-219. https://doi.org/10.1111/1467-9477.12194.

• Renström, E. A., Bäck, H. & Carroll, R. (2021). Intergroup Threat and Affective Polarization in a Multi-Party System. Journal of Social and Political Psychology, 9(2), 553-576. https://doi.org/10.5964/jspp.7539.

• Renström, E. A., Bäck, H. & Carroll, R. (2022). Protecting the Ingroup? Authoritarianism, Immigration Attitudes, and Affective Polarization. Frontiers in Political Science, 4(July), 1-12. https://doi.org/10.3389/fpos.2022.919236.

• Renström, E. A., Bäck, H. & Carroll, R. (2023). Threats, Emotions, and Affective Polarization. Political Psychology, 44(6), 1337-1366. https://doi.org/10.1111/pops.12899.

• Ridge, H. M. (2020). Enemy Mine: Negative Partisanship and Satisfaction with Democracy. Political Behavior, 44, 1271-1295. https://doi.org/10.1007/s11109-020-09658-7.

• Ridge, H. M. (2021). Just Like the Others: Party Differences, Perception, and Satisfaction with Democracy. Party Politics, 28(3). https://doi.org/10.1177/1354068820985193.

• Robinson, W. P. (1996). Social Groups and Identities: Developing the Legacy of Henri Tajfel. Butterworth Heinemann.

• Röllicke, L. (2023). Polarisation, Identity and Affect - Conceptualising Affective Polarisation in Multi-Party Systems. Electoral Studies, 85(August). https://doi.org/10.1016/j.electstud.2023.102655.

• Russo, L., Turkenburg, E., Harteveld, E. & Heckhausen, A. (2023). This Is Not US: Measuring Polarization in Multiparty Systems. A Quasi-Replication Study. Italian Journal of Electoral Studies. https://doi.org/10.36253/qoe-14034.

• Sherif, M., Harvey, O. J., White, B. J., William, R. H. & Sherif, C. W. (1988). The Robbers Cave Experiment: Intergroup Conflict and Cooperation. Wesleyan University Press.

• Simas, E. N., Clifford, S. & Kirkland, J. H. (2020). How Empathic Concern Fuels Political Polarization. American Political Science Review, 114(1), 258-269. https://doi.org/10.1017/S0003055419000534

• Tajfel, H. (1970). Experiments in Intergroup Discrimination. Scientific American, 223(5), 96-103.

• Tajfel, H., Billig, M. G., Bundy, R. P. & Flament, C. (1971). Social Categorization and Intergroup Behaviour. European Journal of Social Psychology, 1(2), 149-178. https://doi.org/10.1002/ejsp.2420010202.

• Tajfel, H. & Turner, J. (1986). The Social Identity Theory of Intergroup Behavior. In S. Worchel & W. Austin (Eds.), Psychology of Intergroup Relations (pp. 7-24). Nelson Hall.

• Tajfel, H. & Turner, J. C. (1979). An Integrative Theory of Intergroup Conflict. In W. G. Austin & S. Worchel (Eds.), The Social Psychology of Intergroup Relations (pp. 94-109). Brooks-Cole.

• Terry, D. J. & Hogg, M. A. (1996). Group Norms and the Attitude-Behavior Relationship: A Role for Group Identification. Personality and Social Psychology Bulletin, 22(8), 776-793. https://doi.org/10.1177/0146167296228002.

• Tichelbaecker, T., Gidron, N., Horne, W. & Adams, J. (2023). What Do We Measure When We Measure Affective Polarization across Countries? Public Opinion Quarterly, 87(3), 803-815. https://doi.org/10.1093/poq/nfad033.

• Torcal, M. & Thomson, Z. A. (2023). Social Trust and Affective Polarization in Spain (2014-19). Electoral Studies, 81(January), 102582. https://doi.org/10.1016/j.electstud.2023.102582.

• van Erkel, P. F. A. & Turkenburg, E. (2022). Delving into the Divide: How Ideological Differences Fuel Out-Party Hostility in a Multi-Party Context. European Political Science Review, 14(3), 1-17. https://doi.org/10.1017/S1755773922000121.

• Verplanken, B., Hofstee, G. & Janssen, H. J. W. (1998). Accessibility of Affective versus Cognitive Components of Attitudes. European Journal of Social Psychology, 28, 23-35. https://doi.org/10.1002/(SICI)1099-0992(199801/02)28:1<23::AID-EJSP843>3.0.CO;2-Z.

• Wagner, M. (2021). Affective Polarization in Multiparty Systems. Electoral Studies, 69, 102199. https://doi.org/10.1016/j.electstud.2020.102199.

• Webster, S. W. (2020). American Rage: How Anger Shapes Our Politics. Cambridge University Press.

• Webster, S. W. & Albertson, B. (2022). Emotion and Politics: Noncognitive Psychological Biases in Public Opinion. Annual Review of Political Science, 25, 401-418. https://doi.org/10.1146/annurev-polisci-051120-105353.

• Westwood, S. J., Iyengar, S., Walgrave, S., Leonisio, R., Miller, L. & Strijbis, O. (2018). The Tie that Divides: Cross-national Evidence of the Primacy of Partyism. European Journal of Political Research, 57(2), 333-354. https://doi.org/10.1111/1475-6765.12228.

Noten

• * This research received funding from the Department of Political Science of Maastricht University and was supported by the Belgian FNRS-FWO EOS project NotLikeUs (EOS project no. 40007494; FWO no. G0H0322N; FNRS no. RG3139). I want to thank Hanna Bäck, Magnus Lindén, and Tim Segerberg for their valuable feedback during the PhD course on political psychology at Lund University. I am also grateful for the constructive comments from Lisa Janssen, Ine Goovaerts, Jonas Lefevere, Artemis Tsoulou-Malakoudi, Chris Butler, and my other colleagues at the M2P research group at Antwerp University, as well as from the two anonymous reviewers.

• 1 Agree-disagree: (1) Because of their worldviews, I could never vote for this party. (2) It is important to me that I am not one of those people who vote for this party.

• 2 Due to mandatory voting in Belgium, it asked whether respondents would vote if elections were not mandatory.

• 3 According to paired t-tests, these differences are significant (p < 0.001).

• 4 According to paired t-tests, all pairs of measurements are significantly different (p < 0.01).

• 5 Separate results for each party system are presented in Appendix D.

• 6 Separate results for each party system are presented in Appendix D.