We label others as straight more often than gay; one reason is a motivated reasoning process that avoids applying stigmatizing labels, because heavy consequences can befall in incorrect gay categorizations

Alt, N. P., Lick, D. J., & Johnson, K. L. (2020). The straight categorization bias: A motivated and altruistic reasoning account. Journal of Personality and Social Psychology, Jan 2020. https://doi.org/10.1037/pspi0000232

Abstract: For 70 years, the field of social perception has concluded that perceivers can determine others’ social category memberships with remarkable accuracy. However, it has become increasingly clear that accuracy is only part of the story, as social category judgments are often systematically biased toward one category over another. For example, when categorizing sexual orientation, perceivers label others as straight more often than gay. This straight categorization bias is reliable, has an effect size larger than that for accuracy, and is not exclusively driven by the low base rate of sexual minorities in the population, yet we know little about its proximal causes. Here, we argue that one facet of this bias is a motivated reasoning process that avoids applying stigmatizing labels to unknown others. Specifically, we propose that perceivers ascribe heavy consequences to incorrect gay categorizations, compelling them to gather and integrate available information in a manner that favors straight categorizations. Studies 1 and 2 tested the dynamic nature of the bias, exploring decision ambivalence and the real-time accrual of visible evidence about a target throughout the perceptual process using mouse-tracking and diffusion modeling. Studies 3–5 tested motivational determinants for the bias, revealing that perceivers associate high costs with incorrect gay categorizations because those errors put other people in harm’s way. Studies 6–9 tested the cognitive mechanisms perceivers engage as they search for information that allows them to avoid costly decision errors. Collectively, these studies provide a new framework for understanding a well-documented but poorly understood response bias in social categorization.