W S Miles1, V Shaw, D Risucci. 1. Department of Surgery, Carolinas Medical Center, P.O. Box 32861, Charlotte, NC 28232-2861, USA. will.miles@carolinashealthcare.org
Abstract
BACKGROUND:Interview assessments of surgical residency candidates may be biased by prior knowledge of objective data. METHODS: Each candidate (site 1: n = 88; site 2: n = 44) underwent two interviews, one by faculty members informed only of a candidate's medical school, the second with prior knowledge of the complete application. Interviewers (site 1: n = 28; site 2: n = 14) independently rated candidates overall and on nine qualitative characteristics. RESULTS: At site 1 only, overall ratings were significantly more favorable for unblinded than blinded interviews (23.0 +/- 17.7 versus 32.6 +/- 23.1, P < 0.01). Blinded and unblinded overall ratings correlated -0.01 (P = 0.90) and 0.31 (P = 0.05) at sites 1 and 2, respectively. At site 1 only, overall ratings correlated significantly with USMLE scores, but in opposite directions for blinded (r = 0.32, P = 0.003) versus unblinded interviews (r = -0.32, P = 0.003). CONCLUSION:Interview assessments may be influenced by objective data, and faculty and program variables. The value of blinded interviewing may vary as a function of individual program characteristics.
RCT Entities:
BACKGROUND: Interview assessments of surgical residency candidates may be biased by prior knowledge of objective data. METHODS: Each candidate (site 1: n = 88; site 2: n = 44) underwent two interviews, one by faculty members informed only of a candidate's medical school, the second with prior knowledge of the complete application. Interviewers (site 1: n = 28; site 2: n = 14) independently rated candidates overall and on nine qualitative characteristics. RESULTS: At site 1 only, overall ratings were significantly more favorable for unblinded than blinded interviews (23.0 +/- 17.7 versus 32.6 +/- 23.1, P < 0.01). Blinded and unblinded overall ratings correlated -0.01 (P = 0.90) and 0.31 (P = 0.05) at sites 1 and 2, respectively. At site 1 only, overall ratings correlated significantly with USMLE scores, but in opposite directions for blinded (r = 0.32, P = 0.003) versus unblinded interviews (r = -0.32, P = 0.003). CONCLUSION: Interview assessments may be influenced by objective data, and faculty and program variables. The value of blinded interviewing may vary as a function of individual program characteristics.