Retention at Risk: Reducing Student Attrition at a Public University

An institutional research project using longitudinal data to quantitatively diagnose when and why students leave — helping a college secure future funding under a new retention-based model.

📌PROJECT SCOPE

Client: Texas Tech University Dean’s Office
Timeframe: 2 weeks (part-time)
My Role: Quantitative Researcher, Data Analyst, Presenter
Team: Coy Callison (Graduate Dean of CoMC), Ben Alfonsin
Methods: Longitudinal Analysis, Comparative Retention Analysis, Logistic Regression Modeling
Tools: SPSS, Excel, PowerPoint

🪟 Project Overview

🚀 CLIENT KICKOFF
Partnering with the Texas Tech University Dean’s Office, I investigated eight years of student data to uncover patterns of attrition and transfer. This data would influence how much funding the college received under TTU’s upcoming retention-based model.

🔎 OBJECTIVES

Identify when and why students leave the College of Media and Communication (CoMC).
Analyze academic "danger zones" in the timeline.
Predict future losses using statistical modeling.
Provide actionable insights before funding changes roll out.

✏️ NOTES
Texas Tech was shifting from enrollment-based to retention-based funding. This raised the stakes for understanding student loss — every dropout represented lost revenue.
The college needed fast answers and a clear plan.

👩🏻‍🔬 Methodology

Data Source: Academic records from 2014-2022

Design: Retrospective, longitudinal analysis

Key Measures: Dropout rate, transfer destinations, loss timing

Inspired by statistician Abraham Wald war plane analysis— we must study losses, not just successes.

This image illustrates survivorship bias: the danger of basing decisions only on visible data.

My analysis focused on the “invisible” losses — the students who left without a trace.

🤿 Deep Dive

1️⃣ DATA COLLECTION & FILTERING

We analyzed eight years of student data, identifying when students left and whether they transferred or dropped out. No students were contacted; the work relied solely on existing academic records.

2️⃣ WHEN DO DROPOUTS OCCUR?

Using Kaplan-Meier survival estimates, we found that most attrition occurs between sophomore and junior years — with notable tapering by senior year. This pointed to specific semesters where intervention could be most effective.

3️⃣ WHERE ARE THEY GOING?

While some students transferred to other programs (like Business), over 50% of student losses had no known destination. These “lost” students presented a critical blind spot.

4️⃣ HOW DOES CoMC STACK UP?

Despite the challenges, the College of Media & Communication retained more students than many larger colleges. Only Agriculture had stronger retention metrics, placing CoMC in a promising position for funding.

5️⃣ HOW SHOULD THE COLLEGE ACT ON THIS?

I built a 31-question Qualtrics survey the college could use to reach former students and gather further data on attrition — complete with sample questions, logic flow, and suggested outreach strategies. I also developed a budget proposal and outlined best practices for student intervention programs. I recommended that the college use survey results to fuel a logistic regression model in future analysis — enabling them to identify high-risk students and implement support before losses occur.

💭 Reflection

Industry Takeaways

Funding systems shape strategy. When retention equals revenue, user journeys must be continuously supported — not just started well.
Data storytelling matters. Findings were presented to stakeholders not just with stats, but also through metaphors and framing.
Analyzing losses can be just as important as celebrating wins. Like Abraham Wald’s planes, we must read between the lines in our data.

Personal Takeaways

I sharpened my statistical and predictive modeling skills under tight timelines.
I learned to synthesize organizational data into clear action plans.
I saw how UX-style thinking — empathy, journeys, failure points — applies beyond digital systems and into organizational strategy.