Predictive model for success in algebra : a case study

Abstract

The objectives of this study were to identify noncognitive variables that would help to predict success (pass or fail) in Algebra and use these variables to develop and validate a statistical model to predict the outcome (pass or fail) of Algebra. First year students enrolled in Algebra (n=164) at a private higher education institute were surveyed on their past achievement, educational goals, parents’ educational qualifications. A modified version of a validated noncognitive questionnaire was administered in this study. Significant categorical and continuous noncognitive variables were identified using chi square test of association and test for independent samples respectively. The significant categorical and continuous variables were used as explanatory variables in binary logistic regression with grade in Algebra (pass or fail) as the dichotomous response variable. The best-fitted model was identified using Backward Wald method. The model developed was significant, explained 56.2% the variance of the response variable based on Nagelkerke R2 and correctly classified 81.0% of cases. The errors were random. The significant noncognitive variables were gender, mother possessing a degree or a higher qualification, Realistic Self-Appraisal and the Availability of a Strong Support Person. The variables in the model did not correlate significantly as indicated by tolerance statistics and Variance Inflation Factors. Based on the model, a unit increase in Realistic Self-Appraisal and Availability of a Strong Support Person would increase the odds of passing the Algebra exam by 1.893 and 1.542 respectively. Being a female would increase the odds of passing the exam by .260 times, while the mother possessing a degree or a higher qualification would increase the odds of passing the exam by 8.511 times. Researchers, academics, academic administrators and student support services stand to benefit from this study as noncognitive variables could be used in statistical models to predict success of students from private universities and higher education institutes in Sri Lanka.