Objective: To better understand the impact of USMLE scores and interview scores on the National Resident Matching Program (NRMP) rank of applicants to the residency program at The Children's Hospital of Philadelphia.
Methodology: We evaluated 935 applicants' files from 2000, 2001, and 2002. For each candidate, one interviewer had access to the full application, while the other interviewer was blinded to USMLE scores and grades. Interview scores were generated by both interviewers. Statistical analysis was performed to evaluate relationships between USMLE scores, interview scores, and NRMP rank list number.
Results: There were a wide range of USMLE scores among candidates who interviewed (range 181 to 269, 227.7 +/- 17.1, M +/- standard deviation). USMLE scores were weakly correlated to nonblinded interview scores (r = -0.17), final committee scores (r = -0.26), and NRMP ranking (r = -0.21): P < .0005. Blinded interviews did not correlate with USMLE scores. Both nonblinded and blinded interviews had stronger correlations with NRMP rank list number (r = 0.49, P < .0005 and r = 0.36, P < .0005, respectively). The nonblinded interview accounted for 20.6% of variance in the NRMP rank list order.
Conclusions: Interview scores were the most important variable for candidate ranking on the NRMP list. Furthermore, when interviewers had access to board scores, there was a modest correlation to performance on the USMLE. While interviews may reflect a candidate's personality, they may not effectively measure desired characteristics when access to academic markers is unrestricted. We suggest incorporating blinded interviews into the selection process to give candidates a better opportunity to display communication skills, emotional stability, and "fit" for the program.