Reducing Discrimination: A Bias Versus Noise Perspective

Discrimination can occur when people fail to focus on outcome-relevant information and incorporate irrelevant demographic information into decision-making. The magnitude of discrimination then depends on 1) how many errors are made in judgment and 2) the degree to which errors disproportionately favor one group over another. As a result, discrimination can be reduced through two routes: reducing noise -- lessening the total number of errors but not changing the proportion of remaining errors that favor one group -- or reducing bias -- lessening the proportion of errors that favor one group but not changing the total number of errors made. Eight studies (N = 7,921) investigate how noise and bias rely on distinct psychological mechanisms and are influenced by different interventions. Interventions that removed demographic information not only eliminated bias, but also reduced noise (Studies 1a-1b). Interventions that either decreased (Studies 2a-2c) or increased (Study 3) the time available to evaluators impacted noise but not bias, as did interventions altering motivation to process outcome-relevant information (Study 4). Conversely, an intervention asking participants to avoid favoring a certain group impacted bias but not noise (Study 5). Finally, a novel intervention that both asked participants to avoid favoring a certain group and required them to take more time when making judgments impacted bias and noise simultaneously (Study 5). Efforts to reduce discrimination will be well-served by understanding how interventions impact bias, noise, or both.