Resident self-other assessor agreement: influence of assessor, competency, and performance level.

OBJECTIVES: To review the literature on self-assessment in the context of resident performance and to determine the correlation between self-assessment across competencies in high- and low-performing residents and assessments performed by raters from a variety of professional roles (peers, nurses, and faculty).

DESIGN: Retrospective analysis of prospectively collected anonymous self-assessment and multiprofessional (360) performance assessments by competency and overall.

SETTING: University-based academic general surgical program.

PARTICIPANTS: Sixty-two residents rotating in general surgery.

MAIN OUTCOME MEASURES: Mean difference for each self-assessment dyad (self-peer, self-nurse, and self-attending physician) by resident performance quartile, adjusted for measurement error, correlation coefficients, and summed differences across all competencies.

RESULTS: Irrespective of self-other dyad, residents asked to rate their global performance overestimated their skills. Residents in the upper quartile underestimated their specific skills while those in the lowest-performing quartile overestimated their abilities when compared with faculty, peers, and especially nurse raters. Moreover, overestimation was greatest in competencies related to interpersonal skills, communication, teamwork, and professionalism.

CONCLUSIONS: Rater, level of performance, and the competency being assessed all influence the comparison of the resident's self-assessment and those of other raters. Self-assessment of competencies related to behavior may be inaccurate when compared with raters from various professions. Residents in the lowest-performing quartile are least able to identify their weakness. These data have important implications for residents, program directors, and the public and suggest that strategies that help the lowest-performing residents recognize areas in need of improvement are needed.