←Back

375 – Negative Results: They're Essential!

Peer Review and Publication of Negative Results: Encouraging Rigor, Reproducibility, and the Integrity of the Scientific Record

at Georgetown University Medical Center

Reproducibility and rigor are currently surging as features to be assessed in publications and grant proposals submitted for peer review. Publication bias, resulting from the submission or acceptance of submissions that report "statistically significant" results - and no other reports - limits both rigour and reproducibility. Negative results are essential contributions to the research record in every scientific discipline, but not if they are badly done, incompletely reported, or otherwise ineffectively reviewed. Effective reviewing of negative results can be complicated by the inaccurate perception that studies presenting "positive" results are actually correct; since many studies presenting "positive" results may or do turn out to be irreproducible, actually-negative results are already a large part of the research record. There are additional impediments to the promotion of effective reviewing of all research that includes analyses, whether qualitative, quantitative, or mixed methods. Statisticians and data scientists are uniquely prepared to provide such evaluations, thereby positively influencing the integrity of the scientific record. The Ethical Guidelines for Statistical Practice urge "the ethical statistician" to apply these principles to what the statistician produces as well as what they review. This paper outlines definitions of "negative results" and highlights key features of the planning, execution, analysis, and write up of studies that should be considered in the writing of informative reviews of submitted manuscripts that describe negative (or positive) results. All quantitative scientists are encouraged to review actively and informatively, and to provide specific input and advice to journal editors on the viability of all research results. By carefully reviewing all submissions, including those with "negative" results, statisticians and data scientists can fulfill their ethical and professional obligations while advocating for the integrity of the scientific record.