First-year residents' performances compared with their medical school class ranks as determined by three ranking systems.

Difficulties encountered with deans' letters have prompted residency directors to rely on performance indicators, including class rank, that indicate the comparative positions of applicants. However, the validity of class rank as a predictor of performance in residency is not well established. The present investigation studied the relationship of class ranks to residents' performances, the systems of class ranking that correlated with residents' performances, and procedures that strengthened the relationship between ranking information and residents' performances. Three systems of class ranking are described. With each system, the authors used the same study group of 124 graduates (classes of 1986 and 1987) from a midwestern medical school with a baccalaureate-M.D. degree program; by 1991, each graduate had completed his or her first year of residency. All three systems were reliable for assignment of class rank. All three systems showed modest correlations with residents' performances. Multiple regression analysis revealed that a weighted combination of clinical performance measures bore the strongest relationship to performances in residency. Thus, the authors conclude that the formal computational ranking of students is not essential for the generation of reliable and valid deans' letters that can predict residents' performances. An in-depth evaluation that includes comparative information based on statistically generated formulas, as well as information based on personal qualities, may be a superior alternative.