Why we should rethink the third-person effect: disentangling bias and earned confidence using behavioral data

Abstract

Although positioned as a cognitive bias, third-person effect research has relied on self-reported difference scores that fail to capture bias appropriately. I use pre-registered and exploratory analyses of three nationally representative surveys (N = 10,004) to examine perceptions of susceptibility to false news and behavioral measures of actual susceptibility. Americans consistently exhibit third-person perception. However, some of this perceptual gap may be “earned.” I show that 62–68% of those exhibiting TPP are in fact less susceptible than average. Accordingly, I construct a performance-derived measure of true overconfidence. I find domain-involvement correlates of TPP tend not to hold for actual overconfidence. I also find significant differences in potential behavioral outcomes suggesting the traditional measure may often reflect genuine differences in self and others’ susceptibility to media, rather than a self-serving bias of presumed invulnerability. These results have important implications for our understanding and measurement of perceptual biases in communication research.