Using Predictive Models to Evaluate Programs and Advance Equity

Western Governor's University's (WGU’s) educational support model is centered on personalized interventions; consequently, we developed a Momentum Indicator model (MOMI) that daily predicts each student’s likelihood of achieving full-time credits. The machine learning model includes features showing students’ prior and current performance and engagement. The MOMI is primarily used to identify students who may need support, track the effectiveness of interventions, and develop faculty. For example, cybersecurity faculty provided personalized weekly outreach to low-MOMI students and found that students were 60% more likely to attend live events and 28% more likely to contact faculty after the intervention. The MOMI also helps us to determine if interventions help students who need help most. Notably, students who identify as members of underserved populations are overrepresented in lower MOMI categories. In 2019, WGU launched a university-wide outreach system that prompted faculty to contact students who completed or failed to complete a task that could affect their progress. We included MOMI as a moderator when testing whether outreach improved student performance and found that outreach—particularly when it was prompt—benefitted lower-MOMI students more than higher-MOMI students. This presentation will describe how and why the MOMI was developed and how we apply it. We will also discuss MOMI’s place in WGU’s mission to increase equitable access and attainment.