Publication Bias in “Red, Rank, and Romance in Women Viewing Men,” by Elliot et al. (2010)

摘要

Elliot et al. (2010) reported multiple experimental findings that the color red modified women’s ratings of attractiveness, sexual desirability, and status of a photographed man. An analysis of the reported statistics of these studies indicates that the experiments lack sufficient power to support these claims. Given the power of the experiments, the probability that the observed 12 findings would all reject the null hypothesis is only .005. Thus, the proper interpretation of the findings is that the studies are contaminated with publication bias. Either some experiments with null findings were not reported or the reported experiments were run improperly in a way that inflated the likelihood of rejecting the null hypothesis. Because of the presence of publication bias, the findings in Elliot et al. (2010) should be considered nonscientific or anecdotal. It remains an open question whether the color red influences women’s ratings of men’s attributes.