Retrospective analysis of the behavioral interview and other preadmission variables to predict licensure examination outcomes in physical therapy.

Evidence supporting the use of personal interviews in admissions decisions for health professions programs is conflicting. This retrospective study was intended to (1) quantify interrater reliability for assessing performance on a particular type of structured interview, the behavioral interview, and (2) examine the ability of multiple preadmission variables, including performance on the behavioral interview, to predict first-time performance on the national physical therapy licensing examination (NPTE). Data from 89 interviewees during the 2006-07 admissions cycle were used to examine inter-rater reliability. Data from 141 students who graduated from 2001 to 2005 were used to examine predictive validity of multiple preadmission variables on NPTE performance, including undergraduate cumulative grade point average (GPA), preprofessional science GPA, performance on the Graduate Record Examination (GRE) (including its analytical, quantitative, and verbal subscales), and performance on the behavioral interview. Inter-rater reliability for assessing interview performance was quantified with an intraclass correlation coefficient (ICC1,1). Receiver operating characteristic (ROC) curves and associated validity indices were used to analyze variables that distinguished graduates who did and did not pass the NPTE on their first attempt (alpha = 0.05). The ICC1,1 for assessing interview performance was 0.749. Performance on the verbal subscale of the GRE (ROC curve area = 0.734, p = 0.007) and behavioral interview (ROC curve area = 0.685, p = 0.034) were statistically significant predictors of NPTE performance. This study provides evidence supporting the contributions of the behavioral interview and verbal subscale of the GRE to predict NPTE performance and assist admissions decisions.