To fail or not to fail? Identifying students at risk by predicting academic success

Student attrition problems are manifest worldwide. Therefore, the identification of students at risk of failure is of major importance. In the current study several cognitive and non-cognitive factors are evaluated in terms of predictive power for failure in the first year of university. Next, a model is proposed that can predict who passes and thus identifies students at risk for academic failure. The sample consists of 461 newly enrolled students in psychology and educational sciences programs. Educational background variables were retrieved from the university database and several cognitive (mathematics skills, reading comprehension and vocabulary knowledge) and non-cognitive (motivation and self-efficacy) measures were administered at the start of the academic year. Logistic regression analysis was used to predict first year academic. Four predictors prove relevant for identification of students at risk of not passing the first year: academic self-efficacy, hours of mathematics instruction in secondary education, results on a basic mathematics test and vocabulary knowledge. A model containing these factors identifies 25% of failing students, while maintaining a specificity of 98%. The practical implications of these results are manifold. First, they help identifying at-risk students. Moreover, they are helpful in designing intervention strategies for those at-risk students. Lastly, they serve as a basis for the development of an online tool that helps future students evaluate their cognitive and non-cognitive abilities in order to choose a major that best suits their potential and background