Authors and reviewers who suffer from confirmatory bias

Confirmatory bias induces overconfidence in the sense that people believe more strongly than they should in their preferred hypotheses. This work describes a Bayesian-based formal model to study the effect of overconfidence about the causes of manuscript rejection due to confirmatory bias in peer review. In addition, we also present an online tool that helps authors to study their beliefs about the causes of rejection. This tool takes the authors’ self-evaluated probability of misinterpretation (i.e. confirmatory bias) and self-evaluated probability of perceiving a review signal correlated with the true cause of rejection and a sequence of review signals perceived as input, and gives a prediction of whether there is overconfidence and wrongness in the author’s belief that bias in peer review caused rejection. We continue to discuss the effect of confirmatory bias in the editor-reviewer relationship in peer review process and show that when the strength of informative signals about the manuscript quality is sufficiently weak and reviewer’s confirmatory bias is sufficiently severe, there is a strong probability that the reviewer would erroneously identify the manuscript quality, making the editor less inclined to offer the potential reviewer any incentive to accept the invitation to review the manuscript. Based on these, we offer a theoretical explanation of current practices adopted to improve the review performance (e.g., desk rejection).