Using Analytics to Transform Decision Making in Higher Education: Patience is a Virtue
Higher education is full of buzzwords, like every sector, and the current buzzwords that get the attention of senior leaders are “Big Data” and “Analytics.” But is this old wine in a new bottle? Not exactly: While the basic principles of data analysis haven’t changed, what is different is the processing speed and the ability to automate, test and revise models that can be used for institutional improvement. Big Data in higher education allows for earlier detection of problems and the development and testing of more effective responses. The trick is to ask the right questions and know how to find the answers. Tackling Big Data requires the combination of computing power, proper capture and alignment of data structures, and the knowledge of appropriate statistical techniques that can answer questions in a valid way. It’s about the ability to search for and recognize patterns in data and requires a broad statistical tool kit.
Smart higher education leaders today want relevant data prior to making major decisions. In my 14 years at SUNY Empire State College, I’ve seen the institution move from a culture of anecdote to a culture of evidence. In my early days at the college, projects were funded based on who made the best case at a budget hearing. Basically, who told the best story or anecdote, rather than what was or could be demonstrated to be most successful. Initiatives moved forward based on what decision makers thought made most sense and hoped would be effective. As it turns out, hope is not a strategy and will not necessarily help a university achieve its goals. Assessment plans were not an expectation to getting a project funded and the following year’s budget hearing for the same division did not include an assessment of the project outcome. Times certainly have changed as accountability has become much more critical in higher education. In the era of scarce resources, investment decisions at the upper levels of administration must be made considering return on investments and university strategic priorities. This is where analytics become critical.
SUNY Empire State College was founded in 1971 to deliver higher education in new and unique ways. There were no classes and students were empowered to design their own degrees in collaboration with a faculty mentor. The assessment of prior college-level learning and liberal transfer policies were mainstays. And while the methods of delivery and policies have evolved over the years, it is no surprise that an institution born of innovation would be using Big Data to address modern problems.
One such problem is minority student retention. According to NCES, Black and Hispanic students nationally graduate at significantly lower rates than do their White and Asian-American counterparts. SUNY Empire State College mirrors national trends in this regard. Our data show about a 10 percent gap in graduation rates between Black students and White students, where the lowest graduation rate is for African American men and the highest is for Caucasian women. Many institutions around the country have developed programs to address this particular achievement gap. For example, the City University of New York (CUNY) system implemented the Black Male Initiative (BMI)—a system-wide set of activities designed to help retain African American male students. Empire State College borrowed some key elements and implemented our own Black Male Initiative, which is largely administered from our locations in New York City. The program consists of at-risk outreach, coaching by peers and faculty, support group meetings, career counseling, academic and social events, and networking opportunities with students and alumni.
So how did we use Big Data to uncover patterns that were actionable to most effectively support our BMI? Fullard, King and Nesler (2016) examined differences between African American male students and their white counterparts, as well as African American male students who graduated against those who did not, comparing thousands of cases over the span of years. While there were minor differences with respect to age, financial aid, socio-economic status, degree levels sought, full-time versus part-time status, and major, there were big differences with respect to behavior in the first term of enrollment at the college. Specifically, African American male students were much less likely to successfully complete all of the courses attempted in their first term than were other students. Examining this finding further, only 6 percent of African American male students who did not complete all their courses in their first term went on to graduate by their sixth year, compared to 49 percent who did complete all their courses in the first term. This basically means that African American men are 15 times more likely to graduate if they successfully complete all their courses in their first term.
Armed with this knowledge, the BMI coordinators began a pilot to specifically target African American men in their first term of study. Early engagement and success appear to be keys to ultimately graduating. For students studying online, our analytics indicated that early posts and log-ins, those occurring within weeks 2 to 4 of a 15-week course, were the critical period. Again, Big Data indicated that early engagement is predictive of successfully completing an online course and cumulative course completion ultimately leads to degree attainment. We have also found that between 2009-10 and 2014-15, the course completion rate for African American men who were regular participants in our BMI was 85.8 percent, while the course completion rate for African American men who were not regular participants in the BMI was 69.5 percent; a difference of 16.3 percentage points. An initial comparison of one-year retention rates between these same groups also has been promising.
So how is this information used at the administrative level? Funding priorities are impacted by the effectiveness of interventions. Where previously we hoped things would be effective, today we use Big Data to manage the institution, support student success and have greater impact. Small pilot projects are often the precursors to larger initiatives and it is much easier to make course corrections based on feedback as it comes in, as opposed to addressing sunk costs later.
Now, is there a danger of using Big Data? Analysis paralysis, or analyzing a data set and asking additional questions to the point where a decision is never made or an action is never taken. Leadership needs to recognize that no data set will ever be perfect, so academic leaders need to work with the best information they have available and constantly check back to see if course corrections are needed. Trust in the analysis is essential, which is why a well-trained and resourced institutional research function will make a big difference.
Ultimately, Big Data provides institutional leaders with the ability to understand their students and their own organizations on far deeper—and often more meaningful—levels. Leveraging this information provides new and exciting opportunities to transform outcomes and support student success, which has a positive impact on the student experience, the appeal of the institution to incoming students, and the bottom line all at the same time. But it’s not as easy as flicking a switch. You need to do the work to reap the rewards.