Attitudinal Polarization and Trimodal Distributions: Measurement Problems and Theoretical Implications

Abstract

Objective . We present trimodality as a distributional pattern that is relevant to conceptual issues of attitudinal polarization but fits nowhere along the modality‐bimodality continuum. Methods . We use simulated univariate distributions and distributions taken from the 1994 General Social Survey to investigate how commonly used measures track (or fail to track) noteworthy shifts in the distributions. Results . The kurtosis and variance are largely insensitive to important distributional differences, most importantly that between bimodality and trimodality. Conclusions . We argue that the lack of attention to trimodality and the insensitivity of commonly used measures of polarization (kurtosis in particular) to that distributional pattern create a blind spot in our thinking about polarization and can lead to basic misinterpretations of extant patterns in public opinion data.