A superhumanization bias involves attribution of qualities that are beyond human to a certain group. Waytz and colleagues reported evidence supporting this bias among White Americans wherein Black targets were perceived as more capable of possessing superhuman qualities than White targets. We sought to better understand the nature of this effect by using different response scales (forced choice vs. Likert) and instruction sets (supporting vs. not supporting existence of superhuman abilities). Results across three studies replicate the superhumanization effect and demonstrate the necessity of several key methodological features; however, under the most realistic survey conditions (i.e. allowing unbiased decisions, being truthful about the existence of such abilities), no significant superhumanization bias emerged. Additionally, in conditions with significant bias, the size of the effect was relatively small, suggesting that this bias may not be as widespread as previously believed; indeed, only a minority of participants showed superhumanization in the predicted direction. Findings support the importance of exploring how arbitrary methodological decisions change inferences about psychological phenomena in the population.
Keywords: racial bias; research methodology; superhumanization bias.