Abstract:
We present a meta-analysis of prospect theory (PT ) parameters, summarizing data from 166 papers reporting 812 estimates. These parameters capture risk-taking propensities, thus holding interest beyond PT. We develop an inverse-variance weighted method that accounts for correlations in PT parameters and imputes missing information on standard errors. The mean patterns align with the stylized facts of diminishing sensitivity towards outcomes and probabilities discussed in PT. Beyond this, the analysis yields several new insights: 1) between-study variation in parameters is vast; 2) heterogeneity is difficult to explain with observable study characteristics; and 3) the strongest predictors are experimental and measurement indicators, revealing systematic violations of procedure invariance. These findings highlight the promise of cognitive accounts of behavior in organizing unexplained variation in risk-taking, which we discuss.