• John A Owen, Gregory F Hayden, and Alfred F Connors.
• Center for the Advancement of Generalist Medicine, University of Virginia School of Medicine, Charlottesville, VA 22908-0728, USA.
• Acad Med. 2002 Apr 1;77(4):344-9.

PurposeTo determine the accuracy of admission committee members' predictions regarding which applicants are likely to become generalists, and to determine which applicant characteristics are used and should be used in making these predictions.MethodThirteen characteristics of each applicant who entered medical school in 1990-1993 and graduated in 1994-1997 were obtained from their applications. Committee members reviewed these characteristics and assigned a probability of each applicant's choosing a generalist career. Just before their graduation, the students were surveyed to ascertain their career plans. The relationships between the characteristics and career predictions were analyzed using regression models. A secondary analysis examined the relationship between the students' stated career preferences at matriculation and career plans at graduation.ResultsThe accuracy of the committee members' predictions was low. Predictions of generalist careers were significantly related to seven applicant characteristics: rural legal residence, gender (women), lower science grades, lower MCAT science scores, lower levels of parents' education, no reported research activity, and higher levels of community service. In contrast, the students' actual generalist career plans at graduation were significantly related only to gender (women) and higher levels of community service. In the secondary analysis, applicants' stated career preferences at matriculation were the strongest predictor of their having generalist career plans at graduation.ConclusionsAdmission committee members often made inaccurate predictions about applicants' career plans. This may be because they based their judgments on applicants' characteristics that were not significantly related to the students' career plans at graduation.

13 keywords...

hide…