This dissertation examined the engagement and motivation of 1,044 engineering students and how these constructs related to students' academic development and persistence in engineering. Engagement was assessed based on co-curricular participation (e.g., students' utilization of resources on campus) and motivation was assessed based on students' self-reported expectancies for success and value for the domain of engineering. I applied machine learning techniques to a rich dataset that includes self-reported indicators, registrar data, and many time points of engagement data from various campus activities (e.g., tutoring, advising). Differential predictors emerged as important in predicting motivation, co-curricular engagement, and persistence. Examination of model performance indicators revealed that second-year predictors of late-third-year engineering expectancy and task-value were most robust than models that included other years' data as predictors. In the prediction of co-curricular engagement, first-year predictors and predictors from throughout all three years yielded the strongest predictive capability of the models tested. Finally, in predicting persistence, models including second-year only indicators, third-year only indicators, or indicators from all three years were equally predictive of persistence. For all models, demographic variables contributed strongly to the prediction of the outcomes. Implications are discussed for educational psychology research and for higher education administration.
Read
