Purpose: Despite their widespread use, medical school admission interviews often are unstructured and lack reliability. This report describes the development of a structured admission interview designed to eliminate bias and provide valid information for selecting medical students, with preliminary information about the interview's reliability and validity.
Method: After screening applications, 490 applicants to a public medical school residency program were interviewed by two faculty members using a structured interview format. Interview scores were compiled and correlated with undergraduate grade-point averages (GPAs); Medical College Admission Test (MCAT) scores; Iowa Evaluation Form (IEF) scores, an in-house evaluation of applicants' noncognitive abilities; and eventual admissions status.
Results: Interrater agreement was good; the percentages of rater pairs whose scores differed by one point or less ranged from 87% to 98%. Scores on the structured interview revealed low to moderate correlations with other admission criteria: 10 (p < 0.05) for cumulative GPA, 0.18 (p < 0.01) for MCAT Biological Science, 0.08 (p > 0.05) MCAT Physical Science, and 0.10 (p < 0.05) MCAT Verbal Reasoning. None of the correlations between the overall interview scores and the IEF scores reached statistical significance (p = 0.05). Higher overall scores on the structured interview did predict a greater likelihood of being accepted into the medical school and the interview score accounted for 20% of the incremental variance in admission status beyond GPA, MCAT, and IEF scores.
Conclusions: The moderate-to-low correlations with other admission criteria suggest that the interview provided information about candidate credentials not obtained from other sources and accounted for a substantial proportion of the variance in admission status. This finding supports the considerable time and resources required to develop a structured interview for medical student admissions. Final judgment on the validity and utility of this interview should be made after follow-up performance data have been obtained and analyzed.