Pilot Project: Predictors of Success in a Physician Assistant Program

Moser, Sharon


The selection of appropriate candidates for a physician assistant program is based on parameters that have traditionally been used but have not been proven to be either reliable or valid. This study attempted to begin that process.


Retrospective data was obtained via a chart review of all Physician Assistant students who completed the program at the University of Detroit Mercy between 2003 and 2006, N= 109. Independent variables, termed “predictors” consisted of gender, age, highest degree earned prior to entering the program, undergraduate GPA, grades on each of six prerequisite courses (nutrition, developmental psychology, medical ethics, microbiology, statistics and advanced physiology), GRE (or MCAT equivalency) and it’s component parts, months of medical or helping experience and interview scores. The end point used (“dependent variable”), was the score obtained on the national certifying exam for physician assistants, called the PANCE.


Using a multiple regression analysis R-square test for explanatory power, age showed an R-square value of 0.0125 and gender 0.0121 providing very little explanatory power. Regression analysis showed very little explanatory power for graduate degree vs. no previous graduate degree. The independent variables with the highest predictive values for PANCE scores included undergraduate GPAs, prerequisite score on advanced physiology (highest) and microbiology, GRE (verbal) combined with GRE(math); including the analytical score decreased predictive power. Two tailed t-tests confirmed their significance at the .01 level. Number of months of experience had virtually zero explanatory power. Interview score provided little predictive power for the endpoint PANCE score.


The most predictive indicators include traditional GPA,GRE, and 2 prerequisite course grades (physiology and microbiology). The interview score has no predictive power for students who were admitted. If the population of students not admitted were included, the above variables would be hypothetically more extreme. Interviews are traditionally included as a chance to test the student for “fit” and introduction. A study examining the seven components of the interview and predictive value of portions of the interview among both admitted and non-admitted students would be an area for further research.