Enhancing the Reliability of the Multiple Mini-Interview for Selecting Prospective Health Care Leaders

Abstract

The David Geffen School of Medicine at UCLA Program in Medical Education (UCLA-PRIME) used a 12-station multiple mini-interview (MMI) circuit to assess applicants. The authors sought to determine the reliability of the MMI, potential bias in scores, and the degree of acceptance by interviewers and applicants. In 2009, 28 interviewers interviewed a cohort of 76 applicants. An anonymous survey assessed interviewers' and applicants' satisfaction with the MMI process and perceived bias. Psychometric properties were determined with generalizability and decision theory. The process was repeated the following year with a new cohort of 78 applicants and minor modifications aimed at improving reliability. The MMI format was well received by both applicants and interviewers. No bias based on gender or disadvantaged status was found. The preliminary reliability of the MMI in 2009 was 0.58-lower than reported in previous studies-but improved in 2010 to 0.71 after an easy station was replaced with a more challenging one and a new scoring rubric was introduced. This interview technique proved to be reliable and was seen as transparent, uniform, and fair. The predictive validity of this process remains to be determined.