Increasing student success through statistical modeling for student retention: Lynchburg College (Virginia)

by

College students have a diverse range of needs to help them persist and graduate. Predictive modeling for student retention allows campuses to identify students and match them with the resources they need.Lynchburg College is a small, private liberal arts institution serving students from diverse geographic regions with a wide range of academic performance (high school GPAs ranging from 2.5 to 4.0). Because of this diversity, Lynchburg College has made it a campuswide priority to understand the needs of incoming students and connect them to the resources they need to succeed.

Since 1998, the campus has used the College Student Inventory (CSI), an assessment for incoming freshmen. The CSI measures the self-reported strengths and challenges for students along with their receptivity to assistance. It has helped Lynchburg identify at-risk students earlier and make more targeted interventions. The student data also provide valuable information for academic advisors.

After more than a decade with the CSI, Lynchburg decided to enhance its retention efforts by adding another component: predictive modeling for student retention with the Student Retention Predictor (SRP). This combines the college’s enrollment data with the CSI data into a powerful statistical model that provides comprehensive student retention analytics. Being able to gather data from multiple points efficiently and effectively has helped the college build on proactive retention strategies already in place.

One key factor that spurred Lynchburg to add this predictive modeling element was the inclusion of a sweeping, integrated online portal called the Retention Data Center. This portal not only makes it very easy for the campus to manage its retention analytics, but also allows it to prioritize first-year students based on their retention scores, run key strategic reports, customize the scope of student records, and provide advisors with key information about individual students or populations.

“With the Student Retention Predictor, we are able to not only see a more comprehensive picture, but we are able to get into the Retention Data Center and do something with the data. It isn’t just there to look at—it is there to use,” says Mari Normyle, assistant dean for academic and career services at the college. This became increasingly crucial for the campus as its enrollments grew over the years, especially after enrolling the largest class in school history in 2011.

With student retention, there are a host of variables that can contribute to a student persisting. Academics, motivations, family background, social attitudes, and finances are just some of the myriad factors that can affect whether students complete their educational goals. But what role do these factors play in retention?

Guiding student retention strategies with relevant data

Predictive modeling provides an answer to that question in two ways. First, the predictive modeling process uses all relevant variables to provide a retention score for each student. Some students receive high scores—they are highly likely to persist. Some receive low scores, indicating that they are unlikely to persist, even with assistance from an institution. Most importantly, predictive modeling identifies and prioritizes the large middle group of students, those who are at-risk but also likely to be responsive to efforts to retain them.

This is what elevated Lynchburg’s retention efforts to a new level. After scoring students through the SRP, Dr. Normyle and her colleagues were able to see at a glance where they could have the greatest impact on retention.

In addition, predictive modeling uncovered which variables were the most predictive of a student persisting at Lynchburg. There were four in particular that were critical to student retention:

  • Student GPA
  • Student’s desire to finish college
  • Student’s academic confidence
  • Student receptivity to assistance

While these individual factors may seem obvious, the model revealed how the interplay of factors could sometimes hide potential retention problems. For example, Lynchburg had long had students with lower high school GPAs go through freshman seminar and college skills courses—a strong, proactive retention practice. However, the predictive modeling process also identified students with acceptable GPAs but who lacked study skills as also being ideal candidates for this initiative as well.

Lynchburg’s predictive model for retention has also provided them with the information for targeted outreach communications outside of the learning communities. The college shared additional SRP information with academic advisors, then looked at students who were a high priority based on their number of risk factors in order to prioritize attention from those advisors. Furthermore, Lynchburg College has used the CSI and SRP as relationship-building tools, providing information that helps academic advisors have targeted and potent conversations with students. These tools have allowed advisors to strengthen the relationships they already have so they can become more effective in their work with students.

Seeing greater student success after statistical modeling

While the college cannot attribute all gains solely to predictive modeling, Lynchburg believes firmly that there is a strong correlation between the predictive modeling process and these very positive gains because of the data predictive modeling provided. The campus has seen these results:

  • Midterm grade average as a class is currently the highest it has been in 10 years.
  • Fall-to-spring retention is 91 percent, a 2 percent increase over the previous year (and an increase of 45 additional students who were retained).
  • One of the at-risk groups achieved 93 percent retention for fall to spring.
  • Male retention is at 92 percent, female retention at 90 percent—the first time that male retention has surpassed female.
  • In another at-risk group (80 students), 14 additional students returned for the spring semester.

“With the addition of the Student Retention Predictor, we have been able to build on the power of the CSI and get a more comprehensive picture for our retention efforts,” Normyle explains. The campus plans to continue using student motivational assessment and predictive modeling so it can intervene with the students who need and want assistance, and keep more students on the path to completion.

Do you have any questions about how Lynchburg College used predictive modeling to strengthen student retention? Curious how you could adapt this process for your campus? E-mail us with your questions and we will share our recommendations.

About the author

A recognized leader in higher education consulting, Noel-Levitz is committed to helping institutions meet their goals for enrollment and student success. During our 40-year history, more than 3,000 campus clients throughout North America and beyond have invited Noel-Levitz to collaborate with them.

Read more about Noel-Levitz’s experience and expertise »

Reach Noel-Levitz by e-mail at ContactUs@noellevitz.com.

Related posts you’ll find helpful:

  1. Five strategies for improving student retention and college completion among second-year college students

Tags: , , , ,

Posted by Noel-Levitz

2 Responses to "Increasing student success through statistical modeling for student retention: Lynchburg College (Virginia)"

Leave a Comment

E-mail Noel-Levitz
Noel-Levitz