Affective Polarization Index

Description

The Affective Polarization Index (API) was developed by Reiljan (2020). It is based on Dalton’s Polarization Index and measures the (weighted) average difference between partisans’ like of their party, including leaners, and their dislike of all other parties. Reiljan computes the API based on the CSES dataset, but in principle it can be applied to other datasets that contain the required items, i.e., like-dislike scores and partisan identification questions. For example, Garzia et al. (2023) applies the measure to The West European Voter Dataset and national election studies from Australia, Canada, New Zealand, and the United States.

Operationalization

Reiljan proceeds in two steps. They first calculate an AP score for each partisan group (i.e., those who at lean toward a party). In a party system with $N$ relevant parties, the relative AP of each party is defined as: $$AP_n = \sum_{m=1;m\neq n}^N [(Like_n - Like_m) \times (\frac{Vote~share_m}{1-Vote~share_n})]$$ where the subscript $n$ denotes the in-party and $m$ denotes the out-party. Vote shares are expressed as decimals (e.g., 10% = 0.1). In a second step, they calculate the weighted average of all relevant parties’ AP score in a given party system as:

$$API = \sum_{n=1}^N (AP_n \times Vote~share_n)$$ Both steps taken together can be expressed as follows: $$API = \sum_{n=1}^N[\sum_{m=1; m \neq n}^N ((Like_n-Like_m) \times (\frac{Vote~share_m}{1 - Vote~share_n}))\times Vote~share_n]$$

Given a like-dislike scale ranging from 0 to 10, the API score can range from –10 to +10. ​

polaR

# Import Data
cses_imd <- polaR_import(source = "cses_imd",
						 path = "path/to/dataset.dta")

# Compute Measure
cses_imd <- api(cses_imd)

Visualization

Use cases

Publications that use this measure: