It's Time for a VP of Data Analytics

David Eubanks, Assistant Vice President for Assessment and Institutional Effectiveness, Furman University

As with so many other things, Star Trek was ahead of its time in featuring a science officer on the ship’s bridge. Such positions in higher education a cabinet-level data analysis expert-are still rare, some sixty years after the Enterprise launched. Part of the reason for this is that data analysis can be seen as an engineering problem: just buy the right software. But while engineering curates a representation of the world, analysis makes meaning of it. Dashboard tools like Tableau or PowerBI can only tell us WHAT the situation is, but cannot tell us WHY or what to do about it.

For example, a dashboard of student GPAs by class will likely show that seniors have higher grades than freshmen, which could lead to the conclusion that admission standards are slipping. An analyst would find more nuanced explanations (e.g. survival effects, where the best-prepared students persist, and selection effects where students pick majors they are likely to succeed in).

Institutional Research offices are the closest many institutions have to a Data Analytics VP.  But without cabinet representation, these offices can be treated like data vending machines instead of internal consultants and may be resourced only at a level sufficient to do mandatory reporting.

What would the VP do?

The main role of the new VP is to cultivate analytical capabilities done in-house, for forecasting, answering WHY questions, designing data intake processes so they are useful and reliable, and building trusted relationships across the university.

For example, your surveys are probably a mess. Surveys of students can be a strategic asset, giving you critical information at low cost, but without central coordination and oversight, surveys become a free-for-all that results in poor quality data distributed in ways that can’t be used strategically, and—worse—waste the resource by over-surveying students, so that they just ignore the requests.

An effective Data Analytics Office can pay for itself by obviating some of the consulting you depend on now. And by facilitating common training and common methods, the VP can create a culture of analysis among mid-level managers that pays off in the quality of data and analysis across the board. Analysis only makes sense in the context of domain knowledge in academic affairs, student life, facilities, business operations, and so on, so creating a cadre of data-able leaders who know what data validation looks like and can interpret regression results is critical. Business intelligence is found in people, not software. This work requires close collaboration with the engineering side of the IT shop to agree on a productive division of responsibilities.

Are you ready for a VP of Data Analytics?

Innovation is fraught with peril, and good ideas easily turn into fads. The integration of a cabinet-level analysis unit within an existing leadership structure is no exception. An analyst is often presenting conclusions that can be perceived as challenging the expertise of the other cabinet members, contradicts a pet project, or questions some deeply-held belief. This psychological minefield is well-mapped in books like Predictably Irrational, Mistakes Were Made (but not by me), Think Again, The Scout Mindset, and others. There’s a difference between needing to be right and wanting to get the right answer, which is a litmus test for how welcoming executives will be to a high-level analytical unit.

One test is to inventory consulting contracts and attempt to have honest conversations about their relative costs and benefits, and how quality is being assessed. Consulting may be done because it’s required (e.g. external audits), or to have someone to blame if something goes wrong, or as a kind of lucky rabbit’s foot, or maybe because the analysis is really good.

The Data VP idea makes the most sense for new presidents who are forming a cabinet and can set the tone from the beginning. However, there’s a lot you can do to enhance the analytical capabilities of the IR office as an intermediate measure, by thinking of that office as a hub of business intelligence that can affect the culture broadly.

Paying for it

In addition to saving on out-sourcing analytics work, you can probably save money on the engineering side too, since data curation and analysis can be done with inexpensive software or use technology that’s already in place for other reasons (SQL + R is magic). The enrollment funnel presents opportunities for making everything better, including revenue. For example, admissions criteria are in effect prediction formulas for student enrollment and retention. How accurate is that formula, and how can it be improved?

Maybe we can pay for a VP by in-sourcing a contract or two. But where do we get mid-level analysts to infuse the operation with? A glance at job listings shows the high demand for data science and concomitant costs. Many universities have an edge, though: if you have a data science program on campus, you have a ready source of real-world data projects for students, who can intern while in school and take jobs you offer before they graduate. I pitch students a three-year path to a data science career, with a professional development schedule that meets their long-term goals. Having all the resources of the university at hand is enormously helpful. This is a different mindset for hiring: rather than looking for a year or two of experience in the registrar’s office with a bonus if they can use Excel, it’s the other way around: we want coding experience with the ability to learn any domain rapidly.

Conclusions

Higher education is rich in opportunities for data analysis to improve operations. Figuring out how to take advantage of those opportunities will be a challenge that institutions must face sooner or later. A science officer on the bridge may be what you need.

Check Out This : Anacapa Engineering and Design

Business Intelligence

inventory