Are financial conflicts of interest leading to reporting error in academic literature ?

Abstract

This thesis examines whether financial conflicts of interest (FCOI) contribute to reporting errors in academic finance research. To this end, we analyze a dataset that flags statistical tests for potential errors. Probit regressions are employed to test the relationship between COI disclosures and the likelihood of errors. Descriptive comparisons showed nearly identical error rates for studies with and without financial conflicts of interest. Baseline probit regressions suggested a small positive association between FCOIs and strong reporting errors, though this effect weakened once inverse class weighting was applied to account for the rarity of errors. The signal proved fragile across alternative specifications, raising doubts about its robustness. The findings therefore provide no strong support for the hypothesis that FCOIs are associated with reporting errors. Instead, structural factors such as larger author teams and publication in higher-ranked journals appeared more consistently linked to lower error rates. Overall, the findings suggest that FCOI does not systematically affect reporting reliability, while emphasizing the methodological challenges of studying rare events in research integrity.