UTILITY OF COGNITIVE AND NONCOGNITIVE FACTORS IN PREDICTING ACADEMIC STATUS AND CURRICULAR SPECIALIZATION OF BEGINNING ENGINEERING STUDENTS
Abstract
The primary purpose of the present study was the examination of the utility of pre- or early-college cognitive and noncognitive factors in predicting later academic status and curricular specialization of students who had begun college as engineering students. Also of major concern were effects of different sample representations or of nonnormally distributed measures upon differentiation results. Cognitive data included SAT scores, high school rank, and average grades in math, science, and English. Noncognitive data included the students' sex, socioeconomic status (SES) measures, and interest scores from the Strong-Campbell Interest Inventory (SCII) and the Purdue Interest Questionnaire (PIQ). During their first semester in fall 1976, 419 beginning engineering students took the two inventories. They were followed up eight semesters later and classified according to both academic status and specialization field. From this original sample was selected a subsample of 317 students who proportionally (by field) represented the 1975 beginning student population as of eight semesters later. The majority of the 63 factors were statistically distributed nonnormally and were thus normalized. Each factor was examined as normalized and nonnormalized data as well as with the original sample and the modified sample. Single-factor ANOVA was performed on each factor under each of the four conditions, and several discriminant analyses were performed on various sets of the factors. Results indicated that cognitive and noncognitive factors were approximately equally useful in predicting academic status, but certain of the noncognitive factors were much more useful than the cognitive factors in predicting specialization. Of special importance, although specialization is a subgrouping of academic status, many factors which differentiated specializations did not differentiate academic status groups. In addition, SATs were no more useful in differentiating groups than other cognitive factors or even some noncognitive factors. Finally, with differences in group representation or with nonnormally distributed factors, the utility of only a few factors varied appreciably. However, under such conditions the set of factors selected for multi-factor prediction tended to be somewhat different while giving comparable reclassification results.
Degree
Ph.D.
Subject Area
Psychology
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.