Academic dishonesty: prognostic model using Theory of Planned Action, academic self-concept and neutralization

Abstract

INTRODUCTION. Academic dishonesty represents an ethical crisis for higher education that affects educational quality. The literature has employed the Theory of Planned Action to explain dishonesty by the effect of beliefs about cheating, subjective norm and behavioral control over volitional intentionality. But affective elements such as academic self-concept and neutralizing ethical responsibility have yet to be considered. This study proposes an integrative model that evaluates whether the functional relationships between these variables predict dishonest behavior. METHOD. By means of a predictive design, 561 university students were evaluated with a psycho-technical battery. The data were analyzed with Machine Learning models of logistic regression and Random Forest Classifier. RESULTS. The tree model showed better fit, highlighting the role of behavioral intention, neutralization and attitude towards cheating as the most relevant predictors. DISCUSSION. The integration of affective variables and the use of Random Forest allows us to discover significant interactions and predictor effects that are useful for formulating educational policies that go beyond the punitive and focus on prevention rather than correction.