“Affective vs. ideological polarization in a Latin American country : evidence from two survey experiments"

Abstract

Using two online surveys conducted in Chile, we study in-depth polarization and the relationship between ideological and affective polarization in Latin America. We implemented them before two highly polarized elections: the 2021 presidential runoff and a 2022 plebiscite to approve the new Constitution. We embedded a 2x2 experiment in the presidential survey. In one arm, we showed a video highlighting positive, non-ideological biographical information about the opposing candidate, which aimed at reducing their affective polarization (Rogowski and Sutherland, 2016). On the other, a video provided information supporting the respondent’s opposite position regarding taxes, aimed at moderating views on this issue. The plebiscite survey included two experiments. First, we randomly activated (short-lived) affective polarization with an open-ended priming question, based on Simonovits et al. (2022), to assess the effects on ideology and democratic views. Second, we asked respondents their agreement with two ideological statements made by speakers with varying voting preferences that we randomized. Descriptive data from both surveys show that ideological and affective polarization in Chile are moderately correlated, and unlike evidence from the US, left-wing citizens are more polarized, especially in affects. Overall, we find no consistent direct effects of manipulating affective polarization on ideological polarization, nor vice versa. However, when ideological positions are presented as said by a voter of the respondent’s choice, agreement with the stance increases dramatically, and, in cases, more so if affective polarization had been enhanced by the treatment. Thus, while affective polarization does not induce polarization in ideology in the abstract, affective responses to partisanship are large when ideological stances come from a partisan speaker, as it happens in real life. Finally, we also find that affective polarization (i) positively correlate with intended turnout behavior, and (ii) undermines democratic attitudes, making polarization a sensitive matter for a region with weak democratic records.