Date of this Version
Hasan, M. & Khan, B. 2020. Variability in the Effectiveness of Psychological Interventions based on Machine Learning in STEM Education
This manuscript presents a framework to investigate the variability in the effectiveness of psychological interventions supported by Machine Learning (ML) based early-warning systems (EWS) in science, technology, engineering, and mathematics education. It emphasizes the importance of investigating the resulting variability and suggests that effective EWS cannot be designed without a deeper understanding of the variability. The framework uses an ML-based model to predict students’ academic performance early in the semester for a Sophomore-level Computer Science course at a public university in the United States. The students were given psychological interventions by sending their end-of-term performance forecast thrice during the semester. A randomized control trial was designed to determine whether interventions made an overall positive impact on students’ academic performance and whether there was variability in its impact. Results suggested that although interventions improved academic performance, they were not equally effective at different performance levels and that students at the same level reacted differently to these interventions.