Relationships between Attrition and Student Performance in a Large-Enrollment Online Skill Training

Thursday 2:30pm
Oral Session Part of Thursday Oral Session B


Mike Brudzinski, Miami University-Oxford
Mayme Kalmer, Miami University-Oxford
Derreck Gossett, Miami University-Oxford
Michael Hubenthal, EarthScope
Gillian B Haberli, EarthScope Consortium
The Seismology Skill Building Workshop (SSBW) is a 14 week online, asynchronous course designed to develop undergraduate skills in seismology and scientific computing. The SSBW has been an important addition to the geoscience recruiting and up-skilling portfolio, enrolling ~800 students annually since 2020, with high percentages of marginalized students and non-geoscience majors. However, there is significant attrition as the course progresses, highlighting the need to identify participants at risk of abandoning to motivate persistence. Initially, we tested metrics based on student performance on assignments in the learning management system (LMS). Specifically, we measured trends leading up to abandonment for 3 metrics: 1) how long it took a student to finish an assignment (Duration), 2) the score a student received on an assignment (Score), and 3) when a student turned in an assignment relative to the date it was due (Days before Due). For all participants from 2020-2022 that completed at least 10 assignments, we measured trends by aligning on abandonment and then stacking the time series. We determined that the Days before Due metric had the largest correlation with student attrition, with students submitting assignments later as their abandonment grew closer. The Score metric showed a significant correlation as well, but the drop in score was strongest on the last assignment and still only dropped by less than 2%. These findings indicate that performance on assignments can be predictive, but more correlated variables would improve the forecasting. Our next step is to expand to different student metrics, such as the amount of time since a student's last interaction with course material, and the overall number of course interactions. Future work will also focus on positive interventions to improve student retention, for example adding in incentives, like bonuses or a daily interaction streak, to keep the student engaged in the course.