Nudging Students to Success

UMBC's Data Science team, in collaboration with Mathematics, Psychology, and Computer Science, has developed a model to predict the likelihood of success of a student in a course. This model builds on earlier predictive models designed to assign the likelihood (by midterm) that a student receives a D,F, or W in a given course.

However, this is often too late to intervene as the academic deficit can be insurmountable given the time remaining in the semester. By identifying students that may be at risk at 4 weeks and 7 weeks after the start of the term, an intervention can be provided with the goal of enabling students to seek academic support.

Currently, the intervention takes the form of an email signed by the instructor of the participating course. The intervention is designed to nudge the student towards various academic success resources and encourages the student to explore these options.

In this session, DoIT's Robert Carpenter and Len Mancini will break down the nudges sent by demographic group, college, and program, and show how student predictive scores change from week 4 to week 7. Moreover, we will talk about how the importance of engagement features increases as the semester progresses.