Omission bias, individual differences, and normality

Abstract

Omission bias is the preference for harm caused by omissions over equal or lesser harm caused by acts. Recent articles ( Connolly & Reb, 2003; Patt & Zeckhauser, 2000; Tanner & Medin, in press) have raised questions about the generality of this phenomenon and have suggested that the opposite bias (action bias) sometimes exists. Prentice and Koehler (2003) have suggested that omission bias is sometimes confounded with a bias toward what is normal, a bias they find. We review this literature and report new data showing omission bias with appropriate methods, as well as a small normality bias that cannot explain the omission bias. The data suggest that the bias is largely based on the distinction between direct and indirect causation, rather than that between action and inaction as such. We report substantial individual differences: some subjects show action bias. We argue, though, that concern about omission bias is justified if only a substantial minority of people show it.