John Campbell and Diana Oblinger have co-authored an EDUCAUSE paper on academic analytics that anyone with a practical interest in the topic should read. To begin with, it is a model of how to write a paper that addresses multiple institutional stakeholders across very different domains of expertise. It starts with a clear overview of the goals, breaks down the technical and logistical challenges into terms that non-experts can easily understand, lists out the likely questions, benefits, and risks for each stakeholder group, and presents high-level steps to prepare organizations that are looking to take on such an academic analytics project. And on the substance, it presents a balanced and comprehensive picture of the pros and cons of applying data mining techniques to student information in the service of improving educational outcomes.
I particularly appricate the practical goals that the the authors set: increasing retention and graduation rates. I understand these may seem like pedestrian measures that don’t tell us how much students have learned and don’t necessarily improve the quality of the teaching either. But Campbell and Oblinger make a persuasive case that achieving these goals correlates with better outcomes for students, universities and society as a whole. Students who graduate tend to get better pay and better benefits and are more engaged in civic activities than their peers who don’t graduate, particularly in certain minority communities. Universities with higher percentages of students who graduate save more money per-student on recruiting, leaving more money to invest in improving the quality of education. And the society as a whole benefits from having more people who are not dependent on public support programs, who pay more in taxes, and who are more engaged in politics and other community-focused activities. On top of all this, we actually know how to measure retention and graduation rates. I’m concerned that many of the frantic efforts we on college campuses now to quantify student learning will turn out to be wastes of time and money because we don’t really know how to quantify learning in meaningful ways. Worse, if we delude ourselves about how much we’re able to measure, we may end up distorting the system of educational incentives in ways that actually harm students. This is exactly what has happened in K-12 in the United States with No Child Left Behind.
The Campbell/Oblinger approach may not be sexy, but it’s practically and ethically sound while still managing to be ambitious. This paper deserves your attention.