An algorithm for success: Redesigning college based on student data

Technologies like natural language processing have made it possible to analyze the student voice directly from unstructured data such as phone calls, texts and emails, all in real time and at scale.
Angela Baldasare
Angela Baldasare
Dr. Angela Baldasare serves as Deputy Chief of Staff and VP for Strategic Institutional Research and Planning at National University, which is one of the largest nonprofit Minority Serving Institutions in the U.S.

Three years removed the pandemic, higher education institutions have gained a renewed appreciation of the challenges that so many college students have always experienced, from the pressures of juggling coursework with other responsibilities to financial, food and housing insecurity. Against that backdrop, colleges are feeling a heightened sense of urgency to deploy the right interventions and support to ensure more students can succeed.

Fortunately, we live in a golden age of data. Awash with insights into the student experience, colleges know more than ever before about the specific challenges and opportunities facing our students. But even with the wealth of data now at their disposal, institutions are struggling to piece together solutions that will help significantly more students succeed.

Turning this flood of data into tangible action requires questioning long-held beliefs about what works and what doesn’t. It requires rethinking assumptions about who colleges view as being at risk and why. And it requires redesigning the college experience itself.

Sometimes the changes colleges and universities need to make in response to new insights are obvious and straightforward. In other cases, they require further investigation and a greater degree of introspection about institutional choices. In any case, data means very little if it does not result in thoughtful—sometimes bold—action.


More from UB: With new safeguards rule, cybersecurity programs become even more critical for higher ed


Just over a decade ago, nine colleges partnered to increase student completion by removing the myriad pf barriers present along a student’s academic journey. The initiative, known as Completion by Design, created a framework to help colleges identify where exactly students were meeting the greatest obstacles to persistence and completion.

The Loss/Momentum Framework they employed is instructive because it provides an all-encompassing view of the many points throughout the student journey and college experience when students can go off track. It also asks colleges and universities to consider opportunities to build momentum for students, independent of solving for loss. The framework is just one example of how redesigning the learner experience forces institutions to not only discover new factors that contribute to student success but reconsider what success even looks like.

For instance, in one of my former roles at a public university, I was part of a team that worked to closely examine data on students’ course grades and how those were related to persistence and completion. A startling number of students earning Cs or Bs—seemingly solid passing grades—in foundational writing courses were graduating at significantly lower rates than their peers. This challenged everything university leaders thought they knew about potential loss points, as well as which students needed extra support and outreach.

So, the university reimagined the support it offered students, integrating supplemental instruction, asking faculty to make direct referrals to the campus writing center and having advisors underscore the importance of the course. Writing faculty also shared their own research on the relationship between English composition class sizes and student learning outcomes, which resulted in the university adding many more sections of the course. While a significant investment, the changes proved to be a meaningful win for student outcomes.

Colleges and universities have never been in a better position to do this kind of work. Institutions can mine institutional data, surface trends, disaggregate information to examine inequities and build predictive models. More recently, technologies like natural language processing have made it possible to analyze the student voice directly from unstructured data such as phone calls, texts and emails, all in real time and at scale.

The richness of the insights that we are tapping into is only growing, but this unprecedented insight also brings with it new and increasingly complex questions about the student experience, who needs support and how we provide it. It forces a difficult dialogue about the shifts that are necessary for institutional policy and practice. The question ahead is not whether we can diagnose the challenges, but whether we truly have the willingness to change in response to what we learn.

Categories:

Most Popular