Hidden Emotions and Attitudes in Peer Review Reports: Measurement of the Influence of Review Modes on Publication Bias.

This article investigates how different peer review models affect publication bias by analyzing peer review reports from the journal BMC Psychology using text mining techniques, including Latent Dirichlet Allocation (LDA) topic modeling, sentiment analysis, and content analysis. It compares Results-Blind Review (RBR)—a mode functionally equivalent to Registered Reports (RRs) and pre-registration—and Transparent Peer Review (TPR), finding that RBR reviewers focus more on research quality and methodology, exhibit more stable and less positively biased sentiments, and show a more balanced acceptance of positive and negative results than TPR reviewers. The study confirms the presence of publication bias in traditional peer review and suggests that adopting RBR or similar two-stage review processes can effectively mitigate this bias. It further emphasizes the roles of researchers, reviewers, editors, and publishers in perpetuating or reducing publication bias and advocates for the broader adoption of innovative review models to enhance scientific rigor and reproducibility.