PREDICTING WHO WILL CHOOSE A FAMILY MEDICINE RESIDENCY

Purpose. To create and evaluate a screening instrument and a revised i nterview format for use in the selection of family medicine residents at the McGill University Faculty of Medicine. Method. The screening to ol consisted of two sections: an assessment of academic performance (t he TASS) and an evaluation of applicants' generalist versus specialist orientation (the GSSS); each applicant's file was assessed by two rat ers. The revised interview included specific questions and scenarios r elated to family medicine. All three parts were tested on 143 applican ts from outside the region of Quebec in 1994-95. The results on both p arts of the screening tool and the interview were compared with the st udents' first choices of residency and then were compared with the per formances of the 24 accepted into the program as reflected in their fi rst six-month summative evaluation forms. Data were analyzed through s everal statistical methods. Results. Only the GSSS accurately predicte d the applicants' first choices (for agreement between both raters: se nsitivity, 81%; specificity, 70%; accuracy, 78%). No significant assoc iation was found when comparing matching applicants' scores obtained d uring the selection process with their scores on the six-month evaluat ion forms. Conclusion. The GSSS may prove useful as a tool in the revi ew of applicants' files. More studies are needed to reevaluate the use of the interview in the selection process and to help determine wheth er any selection instrument can accurately predict applicants' subsequ ent performances in a residency.