The blog Graphic Sociology, part of the Contexts community of blogs, provides an excellent forum for discussing the visual presentation of information. The blog's author, Flaneuse(a.k.a., Laura Noren), provides examples of the good, the bad, and the ugly in data visualization with a narrative of "what works" and "what needs work" for each graphic.

Yesterday, Flaneuse had a post on obesity trends that originated at the blog Flowing Data. Nathan Yau, the author of Flowing Data posted a challenge to his readers to make an image that answers the question are people getting fatter faster?1 that improves on the following one:

Flowing Data Obesity Trends

Despite the solutions posted at Flowing Data, I actually think that the original graph is not a bad representation of the data; however, it suffers from a few technical problems that I think would be easy to solve. First, the graph does not indicate that the lines are birth cohorts (though Nathan's text does indicate that is what the lines represent). Second, given that they are successive birth cohorts in the study, I think that the colors could have been used more creatively to indicate that they are successive (i.e., the oldest cohort could have been rendered in light gray and the youngest in dark gray with appropriate scaling in between). There are two reasons that I like this graph better than many of the alternatives. First, it follows people over time, which makes a narrative easier to figure out from the image. Second, because what we are interested in is the change in slope of the percent obese across successive cohorts, the original image displays this very well.


  1. I don't agree with the phrasing "getting fatter," but that's what he wrote.