The development of a prediction model for student performance in an introductory college biology course



Journal Title

Journal ISSN

Volume Title



Forty-five pairs of students, matched on the basis of biology achievement test scores, were randomly assigned to two groups. One group was taught using the Personalized System of Instruction (PSI), while the other was taught by the conventional lecture-laboratory method. All students were enrolled in the same introductory biology course and were given the same performance objectives. The study involved students attending Pan American University in Edinburg, Texas, an institution whose student body is nearly eighty percent Mexican-American. Five predictor variables were selected to predict performance in each group. These predictors included past academic achievement, reading ability, preference of instructional method, motivation and sex. These variables were measured utilizing either existing instruments or those developed by the investigator. Each of the two groups was divided into two subgroups at the conclusion of the treatment on the basis of their final exam scores. The subgroups consisted of the upper fifteen students and the lower fifteen students. For each of the five predictors, comparisons were made across treatments utilizing Student's t and chi square for the analysis of data. Neither the upper nor the lower subgroups differed significantly (.05 level of confidence). Null hypotheses asserting no differences between subgroups for each of the five predictors were not rejected. To determine whether the predictors were related to group membership, data were analyzed by multiple discriminant analysis. The null hypothesis stating no relationship was rejected. The only significant relationship was between the predictor variables and membership in performance groups, regardless of treatment. An analysis of the single significant (.05 level of confidence) discriminant function revealed three important predictors, reading ability, motivation and past academic achievement, in order of their standard discriminant weights. The prediction of performance in an introductory biology course, whether taught by an individualized or conventional method, depends on two major types of student characteristics: the cognitive predictors, i.e, reading ability and past academic achievement, and the affective predictors, such as motivation, contribute to the predictive model. Predictive data collected from these two dimensions can be utilized in diagnosis, remediation, and modification of instructional techniques to meet the individual needs of the students. The predictive model developed is of particular value to institutions with large ethnic minority populations, where reading ability often biases standard academic predictors.