Modeling student task persistence in a learning-by-teaching environment
Persistence, a non-cognitive student attribute referring to ones disposition to attain a specific goal despite adversities, is of particular interest and importance because of its relationship to students academic achievement and other individual and societal outcomes. The predictive power of persistence on student success rivals cognitive ability. Thus, developing persistence in students is equally important as their cognitive skills. Despite repeated claims that persistence is a highly valuable skill, studies on student task persistence in computer-based learning environments are limited.The goal of this study is to derive a quantitative model for predicting the incidence of task persistence among students using a learning-by-teaching intelligent tutoring system called SimStudent. Using behavioral features derived from the students the interaction logs, models are built and validated. The analysis of the models revealed that task persistence could be robustly predicted from students interaction logs using machine learning methodologies. Behavioral features related to engagement and self-regulation were found to influence student's persistence in problem-solving tasks. The study also found evidence of the positive relationship between persistence and student achievement and the existence of different behavioral profiles among persistent students.