In higher education, universities have long been viewed as pipelines, preparing students for productive careers in specific fields. But when it comes to understanding how students actually make their way through college, the “pipeline” imagery fails to capture the twists and turns real people often take along the way.
A group of scholars led by René Kizilcec, assistant professor of information science in the Cornell Ann S. Bowers College of Computing and Information Science, is calling for a new, data-informed model for the study of academic progress that leverages the trove of student data colleges and universities already possess. They urge researchers and policymakers to replace the “pipeline” metaphor with “pathways,” an updated, data-centric approach that accounts for the complexity of university curriculums and students’ journeys through them, providing critical information for researchers, university administrators, and students alike.
In a paper published in the journal Science on April 28, professors from nine universities — including Stanford, Columbia, Texas A&M, and the University of Pennsylvania – said this shift toward a more analytical approach would open the "black box" of college to help administrators design effective curriculums and guide the students who navigate them.
“We are building a new science of academic progress that leverages ubiquitous student and course data with computational methods to understand sequences of choices in higher education,” Kizilcec said. “This will enable new ways to understand the choices students make in college, and provide more transparency and advice to students as they make these fateful decisions. It also provides insights for administrators as they make structural and curriculum changes.”
The problem with pipelines
“In science, metaphors guide our understanding of a problem — they shape our approach to observing the world and the way we communicate our findings,” said co-author Mitchell Stevens, a professor of education at Stanford. “The pipeline metaphor has been useful for many years, but it has come to limit our understanding of how academic progress unfolds.”
If students enter college with one major in mind, but then switch, the pipeline metaphor treats that kind of departure as a “leak,” or a loss, rather than an entry onto another route to graduation. What’s more, a pipeline metaphor suggests a lack of agency on students’ part, the authors say, when in reality, students are making decisions throughout their education.
A switch to pathways could also help pave the way for interventions that promote equity, the authors write.
"Pathways science can help demystify the college experience and shed light on the consequences of students’ choices," Kizilcec said. "This can especially benefit students who are first-generation or low-income, who may not have input from people with significant college experience."
Applying new analytical techniques
In conjunction with a new conceptual model, recent developments in computational science make it possible to analyze complex data on academic progress, the authors write.
Currently, colleges and universities have an often untapped trove of student data – grades, demographics, classes that students take or drop, and how long it takes to graduate. This data can be used to understand how students are making choices in a complex system and how the curriculum's structure could be adapted to accommodate student preferences.
The authors call for building a shared analytical framework and infrastructure, including a system for standardizing data across institutions, available as open-source analytic tools. These tools can be shared and applied across schools and university systems.
This work can also lead to better resources for advising students. One such tool is Pathways, a platform that helps students navigate a university’s curriculum to make more informed choices when selecting classes and majors, which was developed previously by Kizilcec’s group, the Future of Learning lab.
“Understanding the consequences of academic choices – picking courses, declaring majors – can be difficult," Kizilcec said. "A new approach that leverages available data along with machine learning and other tools can increase transparency in academic environments to help students make well-informed choices and help universities design curriculums that keep up with the future of work.”
The paper’s other authors are: Rachel B. Baker from the University of Pennsylvania; Elizabeth Bruch from the University of Michigan at Ann Arbor; Kalena E. Cortes from Texas A&M University; Laura T. Hamilton from the University of California at Merced; David Nathan Lang from Western Governors University; Zachary A. Pardos from the University of California at Berkeley, and Marissa E. Thompson from Columbia University.
Adapted from materials provided by the Stanford Graduate School of Education.