Measuring Affective Polarization in Elections: Punjab and Uttar Pradesh

• In Event: The Causes and Consequences of Polarized Politics

Abstract

Affective polarization is arguably the greatest threat to democracies and elections. Between national parliamentary elections in India held every five years, millions vote in local and state assembly elections which have received less attention. We focus on measuring affective polarization to compare the extent of the affective polarization problem in two large northern states that held assembly elections in early 2022. Large RDD samples were interviewed by phone (CATI) over three weeks, six to nine months campaigns began, and the last three weeks of the campaign before the vote in each state: Punjab (first 3 weeks n=5,607 last 3 weeks=15,570) and Uttar Pradesh or UP (first 3 weeks n=18,064, last 3 weeks n=15,136). A single categorial item is used to measure affective polarization that is less costly and less time consuming compared to the standard battery of 0-10 dislike-like scaled items (one for each political party), which is common in election studies in western democracies. What is lost in terms of leveraging those scaled measure in other ways, is gained by authentic reaction to just one question. Parties in each state, larger and smaller, national and regional parties, each have partisans, defined here as those who voted for the party in the last national election and the last state assembly election. Each party also has affectively polarized partisans, a subgroup of partisans who “dislike or even hate another party.” We use multinomial logit regression models to predict vote choice or intention for four political parties in each state, vote choice is between a challenger and the incumbent in each state. The partisan model uses the partisans of each of the four parties as independent variables along with demographics (Age, Gender (Female), and Religion (Hindu, Sikh in Punjab and Hindu, Muslim in UP) to predict the vote, and the affective polaraization model uses affectively polarized partisans of each of four parties. We compare the proportion of partisans and affective partisans for each party in each period, and their importance for vote choice, controlling for all other variables in the models. The odd ratios in the multinomial logit models suggest that SAD and BJP parties benefited more from their affectively polarized partisans than the INC incumbent in Punjab, where the INC lost to the AAP. In Uttar Pradesh, the odds ratios also suggest that BSP, SP and INC, in that order, benefited from strong support from their affective partisans, compared to the BJP incumbent which remained in office. In conclusion we ask: Is it in a party’s interest to encourage or nudge partisans to become affectively polarized going into final weeks of a campaign? More research on the contexts for affective polarization in India is needed.
Our findings on affectively polarized voters support earlier results from a number of studies in the U.S. context.

Authors

• Holli A. Semetko, Emory University

• Kiran Arabaghatta Basavaraj, University College London