Other Polarization Measurements
Polarization: ideological issue affective elite horizontal mass vertical
Data: other ess bes v-dem
Use Cases: @campos_new_2025 @bernaerts_institutional_2023 @hajnal_two_2025 @hobolt_divided_2020 @iyengar_affect_2012 @zumbrunn_unrequited_2025 @levendusky_we_2023 @iyengar_fear_2015 @kiesel_affective_2025 @kitschelt_leftright_1990 @mehlhaff_groupbased_2023 @patkos_measuring_2023 @silva_populist_2018 @traber_groupbased_2023 @westwood_tie_2018 @blattner_does_2023 @chen_polarization_2021
Description
While most studies on political polarization apply one of the widely used measures of polarization documented in this Encyclopedia, some studies develop their own indices and scales, which are not commonly applied beyond their specific research context.
Short Descriptions
Bernaerts et al. (2023) and Hajnal (2025) use Expert Ratings on Polarization in different countries based von V-DEM data.
Campos and Federico (2025) develop an Affective Polarisation Scale (APS) is an index using survey items measuring three dimensions: othering of the outgroup, aversion to the outgroup, moralization of the ingroup.
Hobolt et al. (2020) ; Iyengar et al. (2012) ; Zumbrunn (2025) ; Levendusky and Stecula (2023) employ a Trait-based Measure of affective polarization, in which respondents are either asked to identify the specific positive or negative characteristics they associate with members of their in-group and out-group, or to indicate the extent to which they link these characteristics to each group.
Iyengar and Westwood (2015) use an Implicit Association Test (IAT), an implicit measure of affective polarization, which records reaction times when associating ingroups and outgroups with positive and negative attributes. Differences in response speed between congruent pairings and incongruent pairings provide an indirect indicator of group preference.
Kiesel and Amlani (2025) utilize an Open-Ended, Self-Coded Measure of Affect, where respondents provide a single word describing voters in their own party and in the opposing party, and then code the sentiment of their word choice themselves. This is then used to calculate similarly to feeling thermometers to calculate the difference between in- and outparty.
Kitschelt and Hellemans (1990) measure Citizen-Party-Polarization by comparing citizens’ policy preferences with the positions of party elites across three dimensions: redistribution, sociopolitical governance, and collective identity. Simple regressions are used, where the slope shows whether parties are less, equally, or more differentiated than their voters.
Mehlhaff (2023) use a measure called Clustered Proportional Contribution (CPC), which captures how much of the total variance in data structured by groups (such as political parties) is explained by differences between these groups. The measure captures the proportion of variance that lies between clusters (intergroup heterogeneity) relative to the total variance, ranging from 0 (no polarization) to 1 (maximum polarization).
Patkós (2023) measures partisan polarization using a Partisan Polarization Index (PPI) by dividing the mean satisfaction of government supporters by the mean satisfaction of opposition supporters.
Silva (2018) measure polarization using a measure of Kurtosis, which captures the degree of bimodality of citizens’ ideological orientations - measured as self-positioning on different political issues - by indicating the extent to which the middle is emptied out. This avoids issue-based measures, which are not comparable across countries and time
Traber et al. (2023) measure ideological polarization as Party-Group Polarization, defined as the extent to which supporters of the most right-wing party differ from supporters of other parties in terms of their average ideological position and the variance of their ideological distributions.
Westwood et al. (2018) ; Blattner and Koenen (2023) use Bargaining Games that explore cooperation and conflict between groups by measuring participants’ willingness to donate money to individuals from different group affiliations
Chen et al. (2021) Issue polarization is estimated using networks of agreement, which capture whether X users retweet content from members of their in- or out-group. This is used as a proxy for issue polarization where polarization is high when users predominantly retweet in-group members and low when they mainly retweet out-group members.
Use cases
Publications that use one of these measures: