Abstract

Hindsight bias—the tendency to overestimate in hindsight what one knew in foresight—is a robust and pervasive human error. A recent study with Wikipedia articles, however, found evidence for a hindsight bias only for disasters but not for any other event category (e.g., elections). Although this might suggest Wikipedia articles to be less biased than individuals, alternative explanations had not been ruled out. The present study set out to answer this question by comparing individuals’ and Wikipedia’s representation of the very same event in foresight and hindsight. In particular, we made use of a state election and surveyed one part of participants before and after the outcome and had other participants rate the corresponding Wikipedia article versions with regard to the extent to which the article was suggestive of a particular outcome and presented it as foreseeable and inevitable. In line with prior research and our hypotheses, we found a hindsight bias at the individual level but not in Wikipedia articles. Applying Bayesian statistics, there was substantial support for the null hypothesis (i.e., no hindsight bias) in Wikipedia. By controlling for the potential impact of participants’ own hindsight bias on their article ratings we can rule out alternative explanations of our findings. Therefore, our findings are the first to demonstrate Wikipedia’s superiority over individuals when it comes to hindsight bias.