Informative versus flashy visualizations, and growth in Harvard Stat concentration enrollment

by Sergiy Nesterko on December 4th, 2011

Some time ago my advisor Joseph Blitzstein asked me to create a visualization of the numbers of Harvard Statistics concentrators (undergraduate students who major in Statistics). The picture would be used by the chair of the department to illustrate the growth of the program for university officials, so I decided to make it look pretty. The first form that came to my mind was showing the enrollment growth over years using a bar plot.

Starting in 2005, the numbers follow exponential growth, which is a remarkable achievement of the department. We then decided to follow the trend, and extrapolate by adding predicted enrollment numbers for 2011, 12 and 13. At that moment, there was no data for 2011.

To make the plot look better, I removed the thick error bars and made them into lines so that they are not in the center of attention.

At this point, I thought the plot looked good enough, with the growing numbers being prominently displayed by the vivid crimson colored bars. I also made animated versions for all iterations, here is the one for the plot above. But, the plot did not quite display the exponential trend in enrollment numbers. Therefore, we decided to make a more concise plot, with the fitted exponential line explicitly present (the animated version is here).

This plot was taken as the final product, as it was deemed to display the emphasized features of the data the best. It is evident that the statistics concentration enrollment is exploding (which in many respects may be due to my advisor Joe Blitzstein, who is teaching the main starting class of the concentration beginning 2006). But what is the other take home message from the perspective of visualizing data?

Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away. Antoine de Saint-Exupery

With the development of advanced visualization tools such as Processing and d3js, it has become easier than ever to create data driven visuals that look stunningly colorful and artistic. However, creating informative visualizations calls for parsimony in the selection of data features, colors, fonts and graphical elements for the plot. The web is abundant with pieces that are more artistic than informative, which is ok as long as the viewer’s goal is to be entertained with an arrangement of colors and shapes, and possibly to play with them by interacting with the visualization. Therefore, before creating a visualization, one should always think about what it is created for: to convey information, to engage the viewer, or both.

Plots in this post were created with d3js.

Tags: , , ,

One Response to “Informative versus flashy visualizations, and growth in Harvard Stat concentration enrollment”

  1. […] It is with great pleasure that we find as one of the main objectives of Statistics2013 “nurturing Statistics as a profession, especially among young people”, and makes us very proud to be one of the participating groups supporting this and other also important goals. Hopefully, this initiative will contribute to the exponential trend that has been noticed in the interest of students towards this topic (you can find graphical representations of Harvard´s stat concentration enrollment here). […]

Leave a Reply