Using data visualization for educational reform

Posted on 10/31/2011 by


I recently wrote a brief post suggesting that data visualization may, in fact, have a place in education. I think a recent find from BYU may be able to elucidate this point further.

David Wiley, a Professor at Brigham Young, suggests that teachers may want to visualize test grades in an easily understandable form using a new visualization termed a “Waterfall” (see more here). I noted in my previous post that I had been doing a pretty rudimentary form of visualization with my grades and students seemed to be reacting well to it. In this visualization,

The vertical axis represents students’ final grades (higher final grades at the top). The horizontal axis represents time, with each cell representing a day in the semester. Each individual row represents an individual student. Finally, the darkness of the water droplet represents the amount of time that student spent that day completing gradable activities.


Wiley calls the technique a waterfall because it is able to rank students in an intuitive way, which I find visually interesting. Instead of placing grades in a spreadsheet and noting the frequency of letter grades, he uses color intensity to represent student performance.  In this way, an instructor can easily come to understand, “the relationship between time-on-task and academic performance.”

(Waterfall, David Wiley)

You may have to take a leap here, but I think this has some real-world potential, but in a different way from what I initially suggested.  At this point in time, the people most concerned about very large trends are administrators, and most of the data analytics happens outside of the classroom. In this sense, the most immediate application of this type of visualization is for internal administrative use.  For administrators and those responsible for determining larger trends in student success, this could be an incredibly useful tool.  For example, my particular college is currently examining its developmental education curriculum, looking at student success and the ability for a unified curriculum to address some ongoing problems in course completion.  A consistent issue in evaluating best practices is determining the relationships between variables–is major a good predictor of completion? What about what about age, or gender, or number of times a student has taken a class–or, as we are finding, a combination of many, many variables? The issue is really that we have too much data.

However, I believe that if this type of visualization were made easily accessible, maybe directly from within the within the school’s gradebook software, teachers could play a more active role in determining the meaning and application of their data in a more formal way.  I think that there is a sort of disconnect between the administrators, who are looking at large trends and attempting to identify bast practices to increase student success, and teachers, who have a very deep understand of their individual student’s performance and potential.  It is the difference between broad and deep data.  This deep data needs to play a greater role in determining policy.  In most situations, then, individual teachers really should play a greater central role in deciding what structural changes should be made in education (to me this, by the way, is more or less what unions do in an ideal setting).

Now, what makes the author’s work so compelling is at it simplifies the data in such a way that teachers can wield it to influence policy at the school, local, or even national levels without sacrificing their teaching time. Wiley’s proposition actually returns to a point I have made a number of times on this blog.  In my opinion, it seems that statistical relationships are, in general, easier to understand and more immediately relatable when viewed rather than described in text. Elsewhere in this blog, I have looked at entire academic papers that were essentially summarized by a single, well-done chart. We should take advantage of the fact that our brains function as pattern recognition machines–we evolved to comprehend complex patterns because heuristical reasoning is incredibly efficient from an evolutionary position (see e.g. here or here; also Axelrod 1984, Hudson 1991, Margolis 1987, Political Psychology 2003), and this type of reasoning should be used to influence educational policy from the grassroots level.

Hopefully work like Wiley’s Waterfall and other tools can serve to strengthen the role of teachers in shaping educational policy, and increase the presence of good, understandable data in the educational reform debate.

Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books

Hudson, Valerie, ed. 1991. Artificial Intelligence and International Politics. Boulder: Westview

Margolis, Howard. 1987. Patterns, Thinking and Cognition: A Theory of Judgement. Chicago: University of Chicago Press.

Political Psychology. 2003. “Special Issue: Neuroscientific Contributions to Political Psychology.”Political Psychology24,4.