“Dear Professor Gelman, I thought you would be interested in these awful graphs I found in the paper today.”

Mike Sances writes:

I thought you would be interested in these awful graphs I found in the paper today.

Sample attached [see above], but the article is full of them.

My reply: This is indeed horrible in so many ways. I hope nobody was looking at that graph on their phone while driving!

At the very least, they could go for the click-through solution.

12 thoughts on ““Dear Professor Gelman, I thought you would be interested in these awful graphs I found in the paper today.”

  1. Cannot unsee a big orange caterpillar with a nose on the right which is pointed downwards, which makes it look a bit sad.

    Don’t worry little one, it’ll all be okay!

  2. Augh! I had to measure the diameters:

    year,crashes,mm
    2016,1715,13
    2007,1631,9
    2008,1637,10
    2009,1925,15
    2010,2510,17
    2011,4879,18
    2012,5337,24
    2013,5260,21
    2014,5925,27
    2015,6318,29
    2016,1644,10

    The mm vs crashes plotting curve looks like a puffy recliner.

  3. Oh, and be sure to click through for the many other graphs. The “traffic crashed by year, time of day” (kernel density? snake-ate-a family-of-ducklings) plot is fascinating.

    • That’s a graph for the ages. I’ve never seen anything like it.

      At first, I thought there was some sort of peak in the time series every hour on the hour. Now *that* would be an interesting result. But it’s not. It is some sort of blanket draped over the vertical bars that are the data points.

  4. If anyone can explain the traffic crashes by day of the week one to me I’d love to hear it. How the hell is the Wednesday bar shorter than Friday (also see Sat. and Mon.)? That one graph shows not just that these figures are a poor way of representing the numbers but it’s entirely possible that the actual figure metrics have nothing to do with the numbers in any of the graphs.

Leave a Reply to justin Cancel reply

Your email address will not be published. Required fields are marked *