I’m a geek, I admit it. I like data, facts, information. Not for itself but for what it teaches us and how it can help us plan and improve social programs, policies and conditions.
Case in point: I just read a fascinating article about an ambitious experiment the University of Texas at Austin is embarking on to increase 4 year graduation rates.
UT’s data team sifted through decades of student data looking for patterns associated with graduation outcomes. They found that educational outcomes reflect factors such as family income and education more than standardized test scores (at UT, only 39% of students whose parents weren’t college graduates graduated on time while 60% of students with college graduate parents did). Armed with this information, UT developed a 14-variable algorithm that prospectively produced probability rankings of graduating in 4 years for its 7200 student incoming class.
The algorithm identified 1200 incoming freshman at risk of not graduating in 4 years (or ever). Those students were invited to participate in an upbeat and enriching “student success program” which provides small classes, peer mentoring, extra tutoring, engaged faculty advisers and community building exercises. Interestingly, the selection criteria are never disclosed to the students, in order to convey confidence in their success rather than fear of their failure.
Additionally, other experimental research at Stanford and elsewhere had shown that under-performing students were held back by their own feelings of not belonging and lack of ability. To address these psychological obstacles, UT provided simple online messaging, as part of the freshman orientation information package, acknowledging these issues as well as the key message that these circumstances and feelings are often temporary. Among “disadvantaged” (black, Latino or first generation to go to college) freshmen, this intervention resulted in marked improvement in first semester credit-completion, an important indicator of future success. The messaging had no impact on the “advantaged” students (who presumably are not plagued by self-doubt).
In light of all the attention on the achievement gap in Minnesota, these efforts are particularly noteworthy. It is encouraging that Texas (!) is devoting the resources to thoughtful approaches based on rigorous research. Obviously, it will take time to see if it works.
Cautionary side note: A little blurb in the Chronicle of Philanthropy caught my attention: “Privacy Worries Close Education-Data Group Supported by Gates and Carnegie Funds”. Apparently similar kinds of efforts in public schools were shut down due to parental concerns about privacy. Big Data gets a bad rap, often for good reason; those who are trying to use evidence to close the achievement gap need to be mindful of this minefield and address it head on.