A performance-based method for early identification of medical students at risk of developing academic problems.

The authors studied three classes at the Albert Einstein College of Medicine and found that students' performances on examinations administered during the third month (November) of medical school were highly predictive of their subsequent performances during the first two years of medical school. The investigation had two components: (1) a retrospective study of the classes of 1988 and 1989, which found that students' November grades from three first-year courses predicted 76% of the variance in the year-one weighted aggregate score and 41% for the year-two score, and (2) a prospective study of the class of 1992, in which three November of year one examination scores of the students in the lowest quarter of the class were highly predictive of their encountering substantial academic problems, with a sensitivity of .77 and a specificity of .99. This performance-based method was found to be more powerful than using the scores on the 1977 version of the Medical College Admission Test or the students' undergraduate grade-point averages, or both, in identifying individual students who were academically at risk.