Is There a Replication Crisis in Medical Education Research?

Acad Med. 2021 Jul 1;96(7):958-963. doi: 10.1097/ACM.0000000000004063.


Scholars are increasingly aware that studies-across many disciplines-cannot be replicated by independent researchers. Here, the authors describe how medical education research may be vulnerable to this "replication crisis," explain how researchers can act together to reduce risks, and discuss the positive steps that can increase confidence in research findings. Medical education research contributes to policy and influences practitioner behavior. Findings that cannot be replicated suggest that the original research was not credible. This risk raises the possibility that unhelpful or even harmful changes to medical education have been implemented as a result of research that appeared defensible but was not. By considering these risk factors, researchers can increase the likelihood that studies are generating credible results. The authors discuss and provide examples of 6 factors that may endanger the replicability of medical education research: (1) small sample sizes, (2) small effect sizes, (3) exploratory designs, (4) flexibility in design choices, analysis strategy, and outcome measures, (5) conflicts of interest, and (6) very active fields with many competing research teams. Importantly, medical education researchers can adopt techniques used successfully elsewhere to improve the rigor of their investigations. Researchers can improve their work through better planning in the development stage, carefully considering design choices, and using sensible data analysis. The wider medical education community can help by encouraging higher levels of collaboration among medical educators, by routinely evaluating existing educational innovations, and by raising the prestige of replication and collaborative medical education research. Medical education journals should adopt new approaches to publishing. As medical education research improves, so too will the quality of medical education and patient care.

MeSH terms

  • Bias
  • Data Analysis
  • Education, Medical / methods*
  • Education, Medical / trends
  • Educational Status
  • Female
  • Health Services Research / methods*
  • Health Services Research / statistics & numerical data
  • Humans
  • Male
  • Patient Care / statistics & numerical data*
  • Patient Care / trends
  • Patient Safety
  • Policy Making
  • Publishing / organization & administration
  • Research Design / trends
  • Research Support as Topic
  • Risk Factors
  • Schools, Medical / organization & administration*