Unraveling the visual effect: Affective polarization and exposure to political information in old and new media

For decades, scholars have analyzed political polarization in terms of ideology—that is, how strongly parties and their supporters diverge on political ideals and policy goals (Sartori, 1976; Dalton, 2008). More recently, research has identified a second type of polarization, one "rooted in distrust, dislike, and contempt […] across party lines" (Gidron et al., 2020: 1). This tendency of voters to increasingly view out-parties negatively and in-parties positively is labelled affective polarization and it is widely considered as a dangerous socio-political development. The proliferation of partisan news outlets and social media are frequently blamed for the current polarized political environment. However, it is far from clear to what extent media exposure actually leads to polarization. Extant research has been limited to the US, focusing in particular on online media consumption (Iyengar et al., 2019; Krupnikov and Ryan, 2022). However, large research gaps remain unexplored. In particular, existing evidence seems to lack: (1) a comparative and longitudinal perspective; (2) a simultaneous consideration of traditional media consumption; and (3) a systematic comparison of the effect of media exposure on different forms of affective polarization (i.e., party and leader affective polarization: see Reiljan et al., 2023). This paper aims at filling these gaps through a large-N comparative analysis including twenty established democracies. Our empirical analyses are based on an original compilation of national election study surveys conducted over the last four decades in established democracies worldwide, and featuring measures of exposure to political information offline/online as well as feelings towards both political parties and their leaders. This constitutes an unprecedented opportunity to empirically assess the role of different media diets on affective polarization towards leader and parties. Our theoretical argument builds on existing works at the crossroad between electoral behavior and communication research, which underline the different electoral consequences stemming from reliance on textual vs. visual forms of political communication (for a review, see: Garzia et al., 2020). According to this strand of research, leaders matter more for individuals with a more television-centric media diet (i.e., individuals more frequently collecting news over television than from reading newspapers) vis-à-vis individuals with a newspaper-centric media diet. Drawing upon these insights, we hypothesize that: H1. Higher exposure to political information corresponds to higher polarization at the individual-level; H2: A predominantly visual (vs. textual) media diet leads to more polarization towards leaders (vs. parties); H3: Reliance on political information online leads to more polarization towards leaders (vs. parties); H4: Individual-level relationships are moderated by system-level factors that are already known to affect the relative importance of the party and leader factors (e.g., regime type, size of the party system, type of media system). From an analytical point of view, multi-level regression techniques, including cross-level interactions, will constitute the backbone of our empirical efforts. Our results have the potential to provide an unprecedented assessment of the role of the various media sources in shaping affective polarization in the Western world and to make sense of its changing patterns across time and space.