News: Data, Research & Evaluation

How School Districts Can Harness Data to Combat Summer Melt

Tuesday, November 5, 2019  
Share |

By guest blogger Dr. Jeremy Raff, Coordinator of College and Career Services at School District of Lancaster, an NCAN member

While the majority of high school students aspire to higher education, the goal of college is often out of reach, especially for low-income and first-generation students. Well-intentioned plans can often go awry, especially in the summer months. Summer melt is not a new phenomenon, though it is gaining additional research attention. National estimates range from 10% to as high as 40%, depending on the research. 

While it may not be unique to us as a district, summer melt hits School District of Lancaster (SDOL) hard. Summer melt erodes an already low percentage of students who plan to attend college after graduation. Recognizing this issue, SDOL recently embarked on a data analysis project to better identify students in need of summer melt support. 

In partnership with Franklin and Marshall College’s Center for Opinion Research, we analyzed our class of 2018 students to determine the key predictors of summer melt. We ran multiple regression analysis on our sample and identified a few key statistically significant indicators: 

  • Senior year GPA.
  • Senior year attendance.
  • Completion of the FAFSA (or listed as N/A due to income or undocumented status).

The three best predictors of college enrollment are not surprising, but the impact of their identification on our college advising process has been dramatic.

Traditionally, SDOL personnel worked off of a checklist with students. If students completed their college application, FAFSA, and confirmed their enrollment deposit, we figured they were set. We would check in over the summer and invite them to events, but we trusted that they were likely to enroll. We previously prioritized outreach to those who had not yet completed these steps. Our new findings showed that many of these students who completed all of the college steps on our checklist were still at risk.

We can now predict with a degree of certainty which of our seniors are most likely to enroll in college in the fall. Armed with this new information, we altered our advising approach. To best organize our efforts, we assigned a risk color to all students. We coded students with a 0-40% likelihood of fall college enrollment red. We coded those with a 40.1-70% likelihood yellow, and we coded blue those upward of 70.1%. 

Students in red and yellow received individualized outreach. Our college and career services staff called, texted, or emailed each student individually, and ran through a summer melt prevention checklist. Our checklist is based on a checklist in the SDP Summer Melt Handbook and contextualized with important dates from our most common institutions. All class of 2019 students received text messages via Signal Vine and were invited to a college sendoff event in August. We also provided the likelihood of enrollment rate to our school counselors. This helped guide them in advising individual students who plan to go to college. 

This whole process is not intended to discourage students but rather to promote conversation. If a student had a low GPA or attendance rate in high school, it’s helpful to have a data point that guides a conversation on why they need to improve in those areas to be successful in college. We won’t know our true impact until we receive fall college enrollment reports via the National Student Clearinghouse. However, we currently find this an incredibly helpful way to organize our work. It identified students who previously were not on our radar and helped us prioritize our outreach. As new data becomes available, we will revise our model and reevaluate what we can do differently.

While your school or organization may not have the data or resources to build your own predictive model, it may not be as hard as you think. Consider reaching out to local colleges and universities, especially the economics, education, statistics, and political science departments. They may be able to help you specify a model like ours. Any organization can benefit from systematically supporting students with low grades, poor attendance, and/or incomplete FAFSA status.