In this ongoing series about the role of both big data and AI in higher education, we’ve looked at the pre-enrollment phase and considered the role of AI in the future of the learning experience. In this segment, I’ll focus upon student retention and success, a subject that continues to gain attention about the cost of higher education, and the risks associated with students beginning college only to leave with no degree but significant debt. Of course, student success and retention are not exclusively economic issues. They are both an opportunity to think about how we can create an environment where more students are making progress, achieving their goals, and reaching the milestone that many highlight, namely the issuing of the diploma.
Before exploring this, I’m compelled to note again that both big data and AI are not values neutral. These can and do amplify some values and muzzle others. AI can judge, restrict, and minimize as much as it can free, empower, and support. For leaders and educators in higher education today, I contend that “the data made me do it” is not a morally defensible position. We must take ownership and responsibility for the systems that we use and create, being the first to seek out, acknowledge, and strive to understand both the affordances and limitations that come with these systems.
We are entering the beginning of an era where a growing number of higher education institutions are embracing early alert and other systems that allow us to identify students who are “at risk.” Sometimes these data trigger specific actions that I suspect are almost always well-meaning, even those using these increasingly complex systems don’t embrace enough nuance to account for the deeply human aspects of rich and vibrant learning communities. Colleges are not factories, but simplistic approaches to student success software risks treating them as such. When we accept the factory comparison, students are not the workers in the factory. They are too often considered the widgets.
This is why it is important to approach this topic with at least three perspectives, each of which offers us important insight.
First, there is promise and potential for bridging achievement gaps, increasing retention, and empowering learners.
Second, it is still important to recognize that both learner agency and learner ownership matter.
Finally, because this is a deeply human endeavor, stories still play an important role in mission-minded education, especially in an increasingly data-driven context.
One of the most noted recent stories about advancements in student success and progress toward graduation is Georgia State University. The GSU academic coaching, early alert system, and portfolio of efforts focused upon student success is collectively producing impressive results. Smaller experiments and interventions around the country are helping to achieve similar, albeit more subtle, results.
Go back even a decade and survey a group of higher education professionals about the best way to increase retention and graduation rates, and you might find that the entire conversation would focus upon stricter admission standards. If you have students who are better prepared, then you get higher graduation rates. It is that simple. Only, that left out the vast majority of people, people who, under the right conditions and with the proper support, are fully capable of achieving high levels of performance and achievement.
This old framework was steeped in outdated conversations about the Bell Curve, fixed and overly simplistic approaches to determining intelligence, and a belief that genetics is the gold-standard guide for what a person can and will achieve. Fortunately, we are finding our way out of this thinking into one where we can see, celebrate, and nurture a much broader array of gifts, talents, and abilities. Some practices in big data and AI have great promise in taking us further into this more open way of thinking. We are entering a wonderful new land where academic success is not limited to the fortunate few, but where it is something that can become increasingly accessible.
Agency and Ownership
Imagine that you identified fifty risk factors related to students dropping out of college. Perhaps you get advanced enough to have predictive analytics that tell you the chance that a given student has of passing a particular course on the basis of past performance. Now what? Do you restrict the student from taking the class? Do you use these data to have a meaningful advising session where you provide the student with good information, and offer support for how the student can increase his or her likelihood of success? Or, do you revisit the design of the course with a series of levels, each level preparing learners for success on the next level?
Notice that having the data about what is likely to happen with a student is only the first step. What you do next will reflect your philosophy of education, as well as your values. Will your decisions reflect a belief in the importance of nurturing learner agency and ownership? Or will you be tempted to skip such matters in pursuit of the fastest or most efficient way to quickly increase retention rates?
Stories are Still Important
While not a higher education example, the following story illustrates an important point for all levels of schooling. Consider this data profile of a high school student. In 2nd and 6th grade he underwent an IQ test that revealed 108. Fifteen years later he completed two different IQ tests that estimated an IQ of somewhere between 130 and 140. How was that student advised after the first test and how might it have impacted the student’s opportunities in school and beyond?
It might be helpful to know that, within twelve months of the first test, that young student had three grandparents die, the student was attacked by a dog resulting in almost two hundred stitches and near death, the student experienced domestic abuse, and the student’s father died. Context matters, and while this might seem an extreme example, the data points on the dashboard are not enough to draw out the larger story. How many future Einsteins have we advised away from physics because the data failed to consider the larger story? How many have we advised away from higher education altogether?
How do we use data and AI to enrich stories and help people live out incredible narratives in their learning journey? How do we avoid simplistic solutions and hiding behind a few numbers or indicators?
Big data and AI are useful in creating the conditions for students to succeed and graduate. Misused or allowed to drive decisions apart from a larger philosophy and set of values, they also risk limiting access and opportunity. As we pursue such commendable outcomes like retention and higher graduation rates, we are wise to recognize the many tricks and traps along the way.