make better figures. and maybe a bar graph.

A speaker (Kurt Smith from the San Diego Sheriff’s Department) at the NIJ conference seemingly talked about using GIS to connect crime mapping with decision making. He actually talked about how to make readable figures that convey information. I wish he could talk to my classes!

GIS is widely used by government officials (in cities and localities all over the US) to map crime. Since decision makers have access to massive amounts of data, they tend to output a lot of data measures. They even measure the measures (And why not? They have so much data!). The San Diego Sheriff’s Department produced a 28 page report every month that was a massive collection of data measures that weren’t well thought out, useful, or meaningful. By the end of the year, they printed out over 13,000 pages of reports that no one really read. So they sat down and read the book Measuring What Matters and started to decide what data measures were important and meaningful. Their result was an 8 page monthly report that conveys more information than the 28 page report that it replaces. Smith also mentioned that not every useful measure had a nice visual, so they formatted those measures in other ways (in this case, it meant not displaying the results on a map and using something like a table instead).

I liked this talk because it highlights a few things that my students don’t want to hear:

• Displaying data and results is an important skill in the real world that can get you ahead in your job.
• Format matters! Students may grumble about how I take points off for unreadable homework, but it counts in the real world.
• Not every figure conveys information. You don’t have to output everything, especially if you don’t have anything to say about it. Every figure should tell a story. The results section of a paper should be built around a few key figures. Don’t tell a bad story that rambles on forever or isn’t interesting.
• A figure doesn’t have to be a figure and a table doesn’t have to be a table. How results are conveyed is just as important as what they convey.

I have had an interest in bar graphs lately. When my three year old daughter switched day cares a few months ago, all the new day care centers I visited were teaching children how to make bar graphs. I rolled my eyes the first time, but then I took interest. Bar graphs are very intuitive, even to children a year or two old. I rarely see bar graphs in OR venues, but I often see bar graphs in non-OR venues. Since bar graphs are so intuitive, why don’t we use them more often?