Perceptions and Predictors of Questionable Research Practices in the Biological Sciences

In spite of growing concerns over research misconduct, there is limited empirical research on the factors that may lead to questionable behavior in research.  This study tested whether faculty were more likely to report that they would engage in specific types of questionable research practices based on environmental and individual factors.  A random sample of 429 biology faculty from 107 universities (40% response) completed a questionnaire that presented research scenarios and asked the probability that they would take the same (questionable) action as the actor in the scenario. They also rated how wrong the action was, the likelihood that harm might result, and the probability of detection and sanctions.  In addition, they rated their perceptions of the fairness of university resource allocations, and other demographic and research-related information.  The scenarios with higher perceived probabilities of the respondent taking the same action described overlooking clinical overbilling by a collaborator (15% mean likelihood), favoritism in employment (14%), and conflicts of interest in peer review (14%).  Faculty also reported a mean likelihood of 11% that they would quietly delete suspicious data provided by a senior colleague in order to avoid a potential disagreement.  Scenarios that had lower probabilities were not reporting a change in sample to the IRB (7%), and reneging on a promise to allow students to be lead authors (8%).  As expected, researchers from R1 institutions spent more time engaged in research (η_p² =.20, p<.001) and reported a higher level of funding support than those from masters universities (η_p² =.05, p<.001).  These differences, however, did not predict their perceptions of the likelihood they would engage in questionable behavior, nor did their perceptions of fairness in their research environments.  The best predictors for perceived likelihood of misconduct were their ratings of harm (for individual scenarios, std betas ranged from -.10 to -.22, p<.05) and wrongness of the actions (β ranged -.25 to -.58, p<.05), and in some cases, the probability of feelings of shame and other consequences (β ranged -.14 to -.26, p<.05) (R²s ranged from .31 to .55, p<.001).  These results suggest that institutional variables may have little effect on decisions to engage in questionable research practices.  Instead training and intervention efforts may need to emphasize the potential harm that can be caused by these practices.