Predicting Academic, Clinical, and Licensure Examination Performance in a Professional (Entry-Level) Master's Degree Program in Physical Therapy

Journal of Physical Therapy Education, Fall 2003 by Thieman, Thomas J, Weddle, Mary L, Moore, Marjorie A

Our admission criteria provided even less predictive power related to the second student performance variable of interest, licensure examination scores. The combination of overall preadmission GPA and coded GRE scores accounted for less than 10% of the variability in NPTE scores. Moreover, the correlation of final MPTGPA and NPTE scores (r=.317) was considerably less than the .648 value reported by Dockter.26 In our data, the correlation between first-year GPA (the dependent measure used by Dockter) and NPTE was still only .367. Thus, using more discriminating first-year grades, we still were able to predict only 13% of the variability in licensure examination scores. Nonetheless, 121 of the 122 students graduated (the missing student voluntarily withdrew in good standing), and their collective first-time passing rate on the NPTR was 92%. Once again, high performance and low variability (X =653.9, SD=36.1) constrained the model's predictive power.

Finally, we were most interested in the ability of our admission data to predict clinical performance. Unfortunately, our selection process does not currently use any admission variables that predict CPI scores. There may be several reasons why our admission variables failed to predict CPI scores. First, there is a certain amount of variability in the complexity and challenge presented to students in the clinical learning environment that could potentially result in either higher or lower CPI scores. Thus, the CPI score could reflect the complexity of the clinic, in addition to the capability of the student. Second, there is variability in the scoring of the CPI by clinical instructors, as observed by the Director of Clinical Education, particularly in the interpretation of the endpoint anchors on the visual analog scale, separating novice and entry-level clinical performance ratings. Additionally, CPI scores may be influenced by the quality of the interpersonal relationship between clinical instructors and students. Third, most of the admission criteria are based on academic measures of cognition, such as grades and (est scores. The admission committee score does encompass noncognitive factors such as leadership, commitment to a career in physical therapy, extent of health care experience, and interpersonal skills as assessed by reference letters, but this composite measure also failed to have predictive power. Our Bndings mirror the difficulty other researchers have found in predicting clinical performance and suggest that further research is needed to explore the factors, whether cognitive or noncognitive, that are critical to clinical competence. The approaches suggested by Mann and Banasiak21 may help identify these critical factors. Examining the correlation between CPI scores and the noncognitive factors measured by Guffey et al17 also may prove helpful. Fourth, clinical skills may emerge over the course of the program in a pattern that largely defies advance specification. This possibility may account for the historic difficulty in finding robust predictors of clinical performance.