Graphical Sensemaking
Monday, January 1st, 2007The New York Times is highlighting an interactive graphic… however, it’s actually thin and gratuitous. The journalistic value can be called into question.

However, this interactive graphic (the above is just a static screen grab–please click through), in the same section, could be very powerful. I’m qualifying it because the variables that are tied to the timeline are rather consistent in relationship. This, of course, says something. It isn’t a limitiation of the design of the graphic however. I think it is a powerful sensemaking tool — it just may be better applied to a different set of data.
Part of me resists using an interactive graphic with this subject matter as a critique of information design — but another part of me says there are very few things more important to make sense of than life and death. So I’ll critique as objectively as possible based on power of the interactive to help make sense of the situation, rather than to editorialize or evangelize beliefs about the war.
Some thoughts on how to make it better:
1. Allow filtering the data through variables. I’m not sure there are enough meaningful variables to do so. Race and age, while important, don’t really help make sense of the what is happening. They are driven by the composition of the force rather than anything else. It is an all-volunteer military… which doesn’t experience the same demographic skews caused by a draft (where conventional wisdom says lower socio-economic groups are disproportionately represented). There is an interesting Army/Marine Corps composition split (which teeters from Army towards Marine Corps and then back) which may hint at deployment timings — but also, when taken in context of the size of force, the Marine Corps probably over-indexed heavily at those peaks. However, on the whole, even with the filtering option I recommend, if no new variables are added, the explanatory power of the graphic doesn’t go up much.
2. Allow for density views rather than just scale. As above, just looking at raw numbers doesn’t tell the whole story. Information density must be considered when mapping populations/geographic distributions. The Anbar province has the largest magnitude of deaths over time, shifting moreso in the recent past. However, what would the circles look like if they represented deaths per population (of the province) or deaths per size (of the province)? What is the relative risk of the area?
3. Provide the complementary data. What is missing from this graphic are the insurgent and civilian data. I’m pretty sure the Times could get their hands on estimates of that data as well. Those variables add more information, more signal, and actually less noise. The data now, since there are so few “views” and “slices”, over emphasize confounders… persuading the casual viewer to form an innaccurate assessment of the situation.
Just a few quick thoughts. I have enjoyed some of the Times interactive news items in the past. The burden for them now, and in the future, however, is to resist USA Todaying into entertaino-graphics or editorializing graphics. The best they can do is objectively help make sense of the information — and this is the hardest type of design — eliminating your self from everything and shaping sense, but not dictating it.



