Improved prediction of success in surgical clerkship through measures of information processing.

A new system for predicting success of surgical student performance has been developed. A test of surgical knowledge, with questions given in the form of analogies, was administered to 16 students in their fourth week of clerkship. While solving test items, students' eye movements and fixations were tracked. By analysis of the recordings, eight scores of information-processing capabilities were derived. The processing scores and conventional predictors of medical school clinical performance were analyzed to determine their power to predict success, defined by ratings given on a 1 to 10 scale by 21 faculty members based on three tests of cognitive knowledge, two performance-based examinations, and faculty reports. The ratings were reliable (generalizability coefficient = 0.72; p less than 0.001). Stepwise regression analysis of all variables selected one MCAT score (science problems) and two information-processing scores to the statistical model that maximally predicted success. Regression coefficient for the science problem subset of the MCAT was 0.42. This was augmented to R2 = 0.77 when information processing variables were included. The increment was significant, F (2, 11) = 9.25; p less than 0.01. A newly developed test, coupled with techniques that made possible the derivation of components of information processing, nearly doubled the power of conventional tests to predict success in surgical clerkship.