The Effects of Embedded Digital Learning Skills Training on Undergraduates' Science and Math Achievement

• In Event: An Integrative and Comparative Analysis of Approaches to Developing Undergraduates’ Learning Skills

Abstract

Programs designed to teach students how to learn are generally successful in increasing their self-reported skillfulness, motivation to learn, and immediate task performance (e.g., Dignath & Büttner, 2008; Hattie, Biggs & Purdie, 1996). To produce these effects, program teachers typically dedicate substantial time and effort in low student-to-teacher ratio settings delivering face-to-face instruction. Training learning skills in this way is common in higher education, even though providing this resource to a potentially large student population requires students to enroll in costly additional credits and they may not transfer the domain-general skills to future disciplinary coursework.
In this project, we aim to overcome obstacles to scaling and transferring skill training by developing a digital program that can be delivered to many students directly in their courses, completed outside of the classroom, and embedded with disciplinary examples or practice opportunities that promote transfer. Early studies conducted with biology students in 200-person lecture courses demonstrate that they were able to complete three Science of Learning to Learn modules (Figure 1, Table 1, see PDF) in two to three hours, and that those who did so made greater use of learning resources than the control group. Further, these students outperformed the control group on a subsequent unit exam (Bernacki, Vosicka, & Utz, 2016). A follow-up study relocated training to the beginning of the semester, replicated main effects on learning behaviors and immediate performance, revealed an effect on delayed performance (i.e., final exam), and showed that performance differences were greater for first-generation students and those from under-represented racial/ethnic groups (Bernacki, Vosicka, & Utz, 2017).
This study examines whether embedding the Science of Learning to Learn curriculum in the learning management system course site for a College Algebra course could have similar effects on students’ behaviors and performance in a critical gateway course for STEM majors at a diverse state university. The high failure rate in this course (~40%) illustrates the critical skill gaps that beset college students and hinder STEM retention. The extension to mathematics further examines how domain-neutral skill training that incorporates many STEM topics as examples might broadly improve academic behaviors and achievement across domains. Eighty students were randomly assigned to conditions and completed the Science of Learning to Learn modules or additional algebra problem sets at the end of the first unit in their math course (i.e., by week 4). The training group significantly outperformed a control group on the remaining two unit exams in the course (Figure 2, ds = .50, .49, see PDF). Significant effects of training were not obtained on a standardized, cumulative final exam administered by the department (d = .11).
Study results show that digital training designed to teach skills that apply across STEM disciplines can improve achievement of college students in both science and math. Additional trace data documenting learning behavior in the textbook companion website will further explore how use of skills taught during training mediates the robust effects on learning outcomes and may reveal particularly potent or potentially scalable aspects of the training program.

Authors

• Matthew L. Bernacki, University of Nevada - Las Vegas

• Nicholas Voorhees, University of Nevada - Las Vegas

• Carryn Bellomo-Warren, University of Nevada - Las Vegas