Are female scientists underrepresented in self-correcting science for honest error?

Abstract

Retractions are among the effective measures to strengthen the self-correction of science and the quality of the literature. When it comes to self-retractions for honest errors, exposure of ones own failures is not a trivial matter for researchers. However, self-correcting data, results and/or conclusions has increasingly been perceived as good research practice, although rewarding such practice challenges traditional models of research assessment. In this context, it is timely to investigate who have self-retracted for honest error in terms of country, field, and gender. We show results on these three factors, focusing on gender, as data is scarce on the representation of female scientists in efforts to set the research record straight. We collected 3,822 retraction records, including research articles, review papers, meta-analyses, letters, and others, from the Retraction Watch Database (2010-2021). We screened the dataset collected for research articles retracted for honest error, excluding retractions by publishers, editors, or third parties, and those mentioning any investigation issues. We then categorized the records according to country, field, and gender, after selecting research articles with a sole corresponding author. Our results show that female scientists account for 25% of self-retractions for honest error, with the highest share for women affiliated with US institutions.