Abstracts Psychology

Use of the college student inventory to predict at-risk student success and persistence at a metropolitan university

by Joneel J. Harris

Using Tinto's longitudinal model of institutional departure as the theoretical basis for this research, the purpose was to determine what extent selected motivational factors measured by the College Student Inventory (CSI) predict academic success and persistence of at_risk students at the University of North Texas (UNT). The study focused on United States citizens and permanent residents entering UNT as at_risk first_time freshmen admitted via individual approval for the fall 1994 semester. The 409 subjects were enrolled in a developmental course titled Personal and Academic Effectiveness where the CSI was administered during the first 2 weeks of class. Selected predictor variables were tested in relation to the separate criterion variables of grade point average and enrolled status during the 2nd and 4th years of the study. Grade point averages and enrollment data for the 1995-96 and 1997-98 academic years were extracted from the student information management system. The research design employed appropriate multiple regressions, multiple correlations, multiple discriminant analyses, and bivariate correlations. Findings confirmed the ability of five CSI factors to predict grade point average (p < .05) of at_risk students over the time frames used in this study. Nine factors predicting enrolled status were also significant at the .05 level; however, results were not meaningful in the 2nd year as factors classified 95% of all subjects as persisters. By the end of the 4th year, the factors were able to predict correct classification of both persisters and nonpersisters approximately 24% better than chance. This research provides support for Tinto's institutional departure model, particularly associated with pre_entry attributes and goals/commitments over time. The CSI is a viable instrument for use with at_risk first_time freshmen at a metropolitan university; however, required enrollment in a developmental course likely confounded the ability of selected variables to meaningfully predict enrolled status during the 2nd year.