Self-assessment: With all its limitations, why are we still measuring and teaching it? Lessons from a scoping review

Self-Assessment (SA) is often assumed to be essential for learning, however, this assumption has been extensively rebutted. Research shows SA has significant limitations, including its lack of correlation with competence.

We undertook a scoping review of SA in medical education (2011-2021) and surprisingly discovered substantial research where SA was erroneously assumed to be a valid measure of successful learning, or a skill needing to be taught. Although the initial intent of our scoping review was to explore where SA is being effectively used to advance lifelong learning, we paused to explore the extent of the problem of its misuse and resulting in wasted research, by examining excluded studies.

From 1151 articles, we identified 207 which ignored the documented limitations of SA. Thirty-nine studies explored SA of learning. This research has limited utility: increasing the accuracy of SA does not improve performance or lifelong learning. One hundred and sixty-eight used SA as an outcome measure to assess a program or intervention, including 63 where self-assessed improvement in knowledge/skills was the sole measure. SA of self-confidence was measured in 62 studies. When confidence was compared with an objective measure of performance, both invariably increased, but confidence did not always align with competence when measured.

Many researchers mistakenly assume the ability to accurately self-assess is essential for learning, so focus on teaching self-calibration rather than evidence-based methods of advancing learning. Other researchers incorrectly suppose that self-reported improvements in knowledge/skills provide evidence of the efficacy of a program/intervention. This is particularly troubling with regards to novices, who may believe that because they self-assess to have improved/gained confidence, they are now competent.

Our findings highlight the significant volume of research being done where SA is misunderstood and/or misused as a measurement. We posit reasons that such research continues to take place and suggest solutions moving forward.