Current policies governing editorial conflicts of interest are ineffective

TL;DR

This study evaluates the effectiveness of current policies on editorial conflicts of interest, revealing limited impact and suggesting that disclosure, rather than recusal, maintains trust and efficiency in the publication process.

Contribution

It provides empirical evidence that existing COI policies are largely ineffective and highlights the importance of transparency over recusal in maintaining trust.

Findings

Papers with editor-author associations are accepted faster.

Current COI policies have limited effectiveness.

Public disclosure of COIs restores trust in scientific publications.

Abstract

Research-active editors face a potential conflict of interest (COI) when handling submissions from authors who share the same affiliation or those who recently collaborated with the editor. Since perception of COIs arising from such editor-author associations may erode trust in science, some policies recommend, and others demand, recusal in such incidents. However, the effectiveness of such measures is unknown to date. To fill this gap, we analyze half a million papers from six publishers who specify the handling editor of each paper. We find numerous papers with editor-author associations, and demonstrate that such papers tend to be accepted faster. A quasi-experimental design exploiting policy changes at PNAS and PLOS reveals the limited effectiveness of current COI policies. A network neural embedding model reveals that requiring editors with potential COIs to recuse may compromise the suitability of the handling editor. Finally, an online survey experiment demonstrates that such COIs influence trust in the paper's finding, but public disclosure eliminates this effect.