It’s well known that colleges struggle to retain a portion of students until graduation. Researchers and college professionals postulate a variety of reasons why students leave a four-year degree program. Some quantitative in nature, others qualitative, the reasons for dropout tend to cluster around the same target group based on psychographic and demographic characterizations, often called the “silver bullet variable.” 1
Our case study is Michigan Tech University (Michigan Tech), located in peaceful Houghton, nestled in Michigan’s Upper Peninsula. Basic demographics do not precisely identify at-risk students as well as often hoped; for example, 14.5% of Pell Grant recipients at Michigan Tech can be categorized at at-risk, but because this segment is so large, it disqualifies it from being actionable, meaning that no real, successful steps can be taken to prevent dropout. Often when this group of students is targeted, colleges do not see results in retention that they expect. The combination of accessible University data with the advancement of data science delivers the ability to take into account many factors that could lead a student to leave their degree. The results of this multi-factored approach may surprise you. It paints a picture clear and steady from which actionable steps towards real retention can be taken.
What we offer at SightLine is a “stained glass theory.” Imagine St. Patrick’s Cathedral in Manhattan, the stained glass windows drenching the interior knave with droplets of sparkling colored light. You walk up to one window and appreciate its beauty, admiring how all of the tiny pieces of glass, the tesserae, work together to cohere into a legible whole. If you were to zoom in to see the glass at a distance of six inches, it would no longer make sense, yet when you take a few, or a hundred steps back, the picture comes into focus, the narrative apparent. We tell this same story with data. The hundreds of tesserae in the stained glass window are the 10s and 100s of factors we can consider with data analysis and predictive models to reveal why a student may leave their undergraduate education. With this level of precision for understanding the bigger picture, we can effectively isolate individuals who will be best served by University professional-led interventions. In other words, we need the big picture to best serve individual students with their educational goals.
Many universities identify two factors for at risk students: first time undergraduates with traditionally at-risk demographics and probationary status. We discovered that GPA thresholds that identify at-risk students are dependent on a host of other factors, for example, how close these students were to graduation. GPA is a non-linear factor when predicting drop out; again, we are best served by considering at once as many variables as possible. Through the research we found that students who participated in activities on campus were more likely to stay to graduation and that some activities were more beneficial than others. Most significantly, student employment played a large role in at-risk student retention. Multi-factor analysis provides the ideal perspective to view the more complex, full narrative at hand, like the ideal vantage point for elaborate stained glass windows at St. Patrick’s Cathedral. The data, like the tesserae, coheres in one glance to provide us with a more complete picture from which impactful action can be taken in stride.
Consideration of combined, multi-faceted factors alone to target at-risk students could create sizable profit for your University. SightLine supports three controllable variables that can be used in addition to student interventions: university employment, GPA in tandem with total term credits taken, and campus student activities. Action needs to be taken, and SightLine will be working this summer to assist Michigan Tech in executing its intervention plan. This multi-factor identification system created by using data in predictive modeling has highlighted a list of specific students for whom intervention has the potential for great impact, and hopeful retention at Michigan Tech.
This stained glass theory for predictive modeling can help you at your University locate and intervene with specific students who are in most need of support. Not only will you be helping students who need it, but you will also be increasing revenue for your school.
1Karen Vignare, “We Need a Strategy -- Not a Silver Bullet -- for Student Success,” Ed Surge, September 8th, 2016, https://www.edsurge.com/news/2016-09-08-we-need-a-strategy-not-a-silver-bullet-for-student-success