Predicting the End of Term Status of Community College General Psychology Students

Abstract

Nine hundred and thirty-five general psychology students from a Western community college were used to construct and validate a statistical model which would predict success or failure in a general psychology course. The researcher performed an exploratory analysis of twenty-eight predictor variables after which he concluded that age, high school grade point average, and composite ACT score were the best set of predictors, R2 = .34; p < .001, of status (i.e., success or failure). A multivariate predication model was fitted and diagnosed for reliability, χ2(8, N = 339) = 11.29, p < .20. The model was confirmed by comparing prediction error rates from two previous semesters of students with that of the model building rates. Overall predictability was 74 percent for both sets of data; however, a failure to predict failure rate ranged between 13 to 14 percent. After modification of the prediction rule, πh < .70, the rate dropped to 7 percent. Suggestions for interpretation and implementation of this model are discussed.