Events

Professor Paul Seabright
Professor of Economics
Toulouse School of Economics

This paper investigates how the characteristics of information presented in a narrative frame influence the ability to make statistical inferences. Three incentivized studies (N = 3958) conducted on U.S. representative samples examined individuals’ ability to infer traits of a population by repeated sampling. In Study 1, participants inferred the proportion of yellow and blue balls in an urn in a narrative-free setting. Studies 2 and 3 had a narrative framing where participants inferred the proportion of successful and failed outcomes among projects funded through a crowdfunding website, with vignettes attributing outcomes to either entrepreneurs’ luck or merit. Individuals made less accurate predictions when the task included statistically irrelevant narrative information, due to biases induced by the characteristics of the frame rather than greater random noise. Moreover, personal traits indicating a particular openness to narratives were associated with lower statistical ability: participants with higher religiosity and greater interest in fiction were less accurate than others – but, surprisingly, this was true whether or not narrative framing was involved.