Addressing researcher fraud: retrospective, real-time, and preventive strategies–including legal points and data management that prevents fraud

Abstract

Researcher fraud is often easy and enticing in academic research, with little risk of detection. Cases of extensive fraud continue to occur. The amount of fraud that goes undetected is unknown and may be substantial. Three strategies for addressing researcher fraud are (a) retrospective investigations after allegations of fraud have been made, (b) sting operations that provide conclusive evidence of fraud as it occurs, and (c) data management practices that prevent the occurrence of fraud. Institutional and regulatory efforts to address researcher fraud have focused almost exclusively on the retrospective strategy. The retrospective approach is subject to controversy due to the limitations of post-hoc evidence in science, the difficulty in establishing who actually committed the fraud in some cases, the application of a legal standard of evidence that is much lower than the usual standards of evidence in science, and the lack of legal expertise by scientists investigating fraud. The retrospective strategy may be reliably effective primarily in cases of extensive, careless fraud. Sting operations can overcome these limitations and controversies, but are not feasible in many situations. Data management practices that are effective at preventing researcher fraud and unintentional errors are well-established in clinical trials regulated by government agencies, but appear to be largely unknown or unimplemented in most academic research. Established data management practices include: archiving secure copies of the raw data, audit trails, restricted access to the data and data collection processes, software validation, quality control checks, blinding, preregistration of data processing and analysis programs, and research audits that directly address fraud. Current discussions about data management in academic research focus on sharing data with little attention to practices that prevent intentional and unintentional errors. A designation or badge such as error-controlled data management could be established to indicate research that was conducted with data management practices that effectively address intentional and unintentional errors.