Predicting MBA no-shows and graduation success with discriminate analysis

This paper uses discriminate analysis to examine five years of MBA admission records in order to separate no-shows from the successful program graduates. The study used traditional numeric data such as age, length of time with current employer, undergraduate GPA and GMAT scores—as well as dummy variables for sex, full- or part-time status, race, the public or private nature of the undergraduate institution, and in-state tuition eligibility. The analysis correctly separated the no-shows with a 94.2 percent classification rate based entirely on the use of dummy variables. Unlike other studies, undergraduate GPAs, GMAT scores, and other numeric variables played no role in the final classification. The results suggest that more attention be given to the use of dummy variables when it comes to predicting the success of MBA program graduates.The authors would like to thank the unfailingly efficient efforts of Geetanjali Goel, who helped retrieve and organize the data examined in this study.