Online Program

A Critical Appraisal of Statistical Practice in Translational Animal Studies (309602)

Keywords: translational research, preclinical research, animal models, statistical misapplication, applied statistics, meta-research

Scientific rigor in translational animal research is notoriously poor. Recent calls to minimize bias have received widespread attention, but a critical appraisal of current statistical practice in animal research is still desperately needed, particularly for work yielding complex or longitudinal data. Thus, we systematically evaluated 260 experiments published 2017-2019 that employed murine models to test motor rehabilitative therapies. We present the prevalence of articles in which authors ignored assumptions of statistical tests or models, failed to address multiplicity or justify sample size, did not account for animals with partial data, or reported too little information to evaluate statistical choices. We determine whether statistical rigor is associated with the type of disease or therapy under investigation, minimization of bias through randomization or blinding, journal impact and criteria, or adherence to transparent reporting recommendations. Implications for the role of statisticians in multi-and inter-disciplinary teams are discussed. Our work provides a foundation for solutions to bring good statistical practice to this critical phase of translational research.