Data Viz Done Right

October 27, 2014

Monday Makeover: The odds of dying from select causes in the USA

Peter Gilks of Paint By Numbers sent me this viz last week for a Monday Makeover.

At first, I thought Peter was trolling me, because this is just so, so bad. My initial thoughts:

  1. Why is there an "any cause" box? We're all going to die so of course the ratio is 1:1.
  2. There are two causes with 1 in 7 odds, yet the boxes are different sizes.
  3. Americans have a 1in 28 chance of dying from Chronic lower respiratory disease, yet that box isn't 1/4 the size of the 1 in 7 boxes.
It only gets worse from there. I would have made this much simpler with a bar chart. Yes, it's boring, but it's more effective.

Download the data here and the Tableau workbook here.


  1. A quick read through made me think Peter had created the original view. I re-read and was glad to see that he had only delivered it to you!

    1. Joshua, I would have totally called him out! Would it be worthy of a Zen inquest?

    2. Hey! I would have done a pie chart for this......

  2. I thought I'd have a go at this... The bar chart is a big improvement from, well, WTF is that original viz, but perhaps the lower rates get lost, save for the text labels. When displaying data with large order-of-magnitude differences, visualisations that use area can work (e.g. bubble chart, tree map) but sometimes you just have to 'segment' the data up. I've seen this done before in stuff like large sums of money: or even huge distances: I 'segmented' this data up into high and low odds and showed them as 1 in 1,000 deaths and 1 in 100,000 deaths. This seemed to be an easier way to comprehend than 1 in 23,350 etc.!/vizhome/Oddsofdying/Oddsofdying

  3. Cheers.... Annoyingly the little icons look great in Desktop but seem to be all pixelated once in the browser. Perhaps they're too small