Data Viz Done Right

November 13, 2012

Need to waste space with a poor chart design? Go to Nielsen for advice.


Nielsen has done some really poor work in the past, but this time they may have outdone themselves.  There are times, many of them, that I wonder what Nielsen is thinking.  I hate donut charts even more than pie charts and donuts inside donuts are an even more profound mistake.

Here’s a chart from their Q3 2012 Consumer Confidence Report.


The report includes some additional facts about change from Q2 2012 (except for North America which they only provide vs. Q3 2011 for some inexplicable reason).  That gives us two interesting and simple data points: (1) current sentiment and (2) change from previous.

I would represent the data as a bar chart colored by % change, with a reference line for Global.

Nielsen Recessionary Sentiment

My version makes it much easier to do a few things:

  1. Compare relative differences between regions (bars work better than slices or donuts)
  2. Quickly see which regions are improving, which regions are declining, and the magnitude of the change (via color)
  3. Compare the regions to the global average

I was considering having another bar for the Global average, but decided to use the reference line because it makes the comparisons significantly simpler and my interpretation from Nielsen’s charts was that you should be comparing to the Global average.

Also, their title is misleading.  They make a blanket statement that “Recessionary Sentiment Grew”, yet it didn’t grow everywhere.  I can only assume that they’re referring to the global average.  My title makes it more clear.

For as long as I’m in this line of work, I will hold out hope that Nielsen will incorporate more best practices into their work.  Until that time, I will continue to leave them comments and suggestions.


  1. There is an interesting effect on these charts though Andy, one that surprised me.

    Whilst we're good at reading length of things, we're also exceptionally good at reading things that look like clock faces. We can tell instantly if something is at 75% on the charts above by comparing it to our understanding of 9 o'clock.

    Not sure if that's a useful observation, but it certainly surprised me.


  2. But unless it's at 3, 6 or 9, it's hard to tell. Also, comparisons are challenging and they take up way too much space.

    But, the biggest thing for me was that they didn't incorporate the % change, which is at least as critical of a part of the story.

  3. Spoken like a Tableau Zen Master! Congrats.

  4. My only basic comment would be reverse your colour

  5. Victor, the color scale is correct. A positive change means sentiment has gotten worse, thus it's red.