Retrospective and prospective hindsight bias: Replications and extensions of Fischhoff (1975) and Slovic and Fischhoff (1977)

Abstract

Hindsight bias refers to the tendency to perceive an event outcome as more probable after being informed of that outcome. We conducted very close replications of two classic experiments of hindsight bias and a conceptual replication testing hindsight bias regarding the perceived replicability of hindsight bias. In Study 1 (N = 890), we replicated Experiment 2 in Fischhoff (1975), and found support for hindsight bias in retrospective judgments (dmean = 0.60). In Study 2 (N = 608), we replicated Experiment 1 in Slovic and Fischhoff (1977), and found support for hindsight bias in prospective judgments (dmean = 0.40). In Study 3 (N = 520) we found strong support for hindsight bias regarding perceived likelihood of our replication of hindsight bias (d = 0.43–1.03). We also included extensions examining surprise, confidence, and task difficulty, yet found mixed evidence with weak to no effects. We concluded support for hindsight bias in both retrospective and prospective judgments, and in evaluations of replication findings, and therefore call for establishing measures to address hindsight bias in valuations of replication work and interpreting research outcomes. All materials, data, and code, were shared on: https://osf.io/nrwpv/.