ERIC Number: ED048329
Record Type: RIE
Publication Date: 1970-Sep
Reference Count: 0
Nontest Variables in the Prediction of Educational Progress.
Schoenfeldt, Lyle F.
Historically, nontest variables have been used more frequently than achievement test scores in the prediction of educational progress, but research in this area has been largely empirical. Recently, however, a theoretical model based on the premise that past behavior is the best predictor of future behavior was formulated. This model was used in a longitudinal study with a sample of 1037 males and 897 females at a southeastern university. Nontest variables were 389 biographical items drawn from a pool of over 2000 items and assembled into a questionnaire. Subgroups were formed on the basis of scores on biographical factors, thus clustering similar profiles, rather than on the item responses. A cross validation study was carried out with a 118-item biographical questionnaire. Two types of educational criteria were used: (1) grade point average and (2) number in a subgroup who were in an arts-science curriculum, number with one or more dean's list, number with one or more probation or dismissal, or number of dropouts. Relationships between these criteria and subgroup membership were univariately and multivariately analyzed. Subgroup structure was stable over a two-year period. The relationship between group membership and selected educational criteria was confirmed, showing that knowledge of a person's past behavior can permit meaningful prediction of educational progress. The female and male biographical factors, subgroup differences on biographical factors and selected criteria and group means are included. (CK)
Publication Type: N/A
Education Level: N/A
Authoring Institution: Georgia Univ., Athens.
Note: From symposium "Use of Nontest Variables in Admission, Selection, and Classification Operations" presented at the Annual Meeting of the American Psychological Association, Miami, Florida, September 1970