1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar

The New York Times—normally a source of clear and interesting infographics—produced the following graphic over the weekend.

NY Times square pie graphic

This is bafflingly awful—it’s Tiger Woods carding a 90. Square pies are an infographic seasoning—they’re cilantro, not steak. Here are a few of the problems with this graphic:

The color choices are bad. The saturations between groups are considerably different. The yellow is highly saturated while the other colors are not. The increased saturation draws your attention to the yellow area, but this is just a category like the others. I’d imagine someone with red-green color blindness would have trouble distinguishing the other colors.

There’s a hole in the center. Presumably this indicates people who didn’t respond to the question, but this is not noted. There are no gridlines in the white section even though the non-responding group should be treated visually like the other groups.

It’s hard to compare the sizes of groups. People are better at comparing lengths than volumes. Mixing length and volumes—some of the of the response categories are arranged linearly, while the inner category is basically a volume (with a hole!)—makes it nearly impossible for people to use their spacial skills to side up the differences. Asking people to compare lines and donuts is like asking whether you prefer the color blue or raw carrots. For the record, I prefer carrots.

If you’re interested in the concept of square pie charts, the place to start is at EagerEyes. If you want to learn how to make them yourself, check out our contest, results, and screencast.

The Times is still a source of great design and inspiration. Here’s another graphic they also produced over the weekend that shows cancer incidence, survival rate, and gender differences in a way that is clear, clean, and concise.

NY Times cancer graphic

Topics:
, ,
  • http://www.supportanalytics.com Tony Rose

    Ughh. This square pie is horrid. Possibly a good example of some folks trying too hard to “spice” up data visualization. I absolutely agree with your feedback regarding this being awful. Kind of like, A for effort, but F for results. I am still debating whether this is better than a regular pie chart. My conclusion is no. A regular pie would have at least shown this a little bit better. The best way would have probably been a bar chart.

  • Jim

    Good analysis. You are right about the color problem. I am (mildly, I believe) red-green color blind and I basically can’t tell the difference between the “return to earth in a different form” and “live in a different place” sections without looking for the thick grid line in between. Even with the line it took me about 5 seconds to find the break.

  • http://juiceanalytics.com Chris Gemignani

    Tony, One of the really bad decisions here is how the pie is filled by spiraling in from the outside. Filling the square pie by filling a row of blocks horizontally until you reach the target would have been a little better.

  • http://www.supportanalytics.com Tony Rose

    Yes, exactly. That spiraling is extremely difficult to follow and adds no value. Using a square pie and shading starting in the lower right for each value would have been better, but would have created five graphs. I think a bar graph would give the readers the same information and would cut down on the comprehension time.

    Your post back in December shows an example:
    http://www.juiceanalytics.com/writing/2006/12/solving-the-pie/

  • Jordan Lund

    My initial reaction towards square pie charts, pixel charts, or whatever you want to call them is that they are horrible and unreadable.

    However now that I’ve had time to think about it, they could work provided there were a style manual for setting them up.

    For example:

    1) Charts must be filled in from left to right from the largest value to the smallest value.

    2) Data labels must be placed in the first column which contains a majority of that value.

    3) It is allowable to complete a column with a smaller value if this will prevent other columns from being broken un-naturally.

    So, taking the NYT data and applying the three rules I just created, you end up with this:

    http://img293.imageshack.us/img293/5534/gridchartco5.jpg

  • http://www.shermandorn.com Sherman Dorn

    Okay, the square-pie nonsense trumps today’s awful bubble-map on attacks in Afghanistan (http://www.nytimes.com/imagepages/2007/08/12/world/20070812_AFGHAN_GRAPHIC.html). Good article, bad infographics, no cookie for Times graphics folks today!

  • Bob

    It’s a visual gag.

    It’s supposed to look like the light at the end of a tunnel, not make it easier to compare the proportion of 4 categories (which even a seven year old can do, even without the aid of a chart, and certainly without complaining about how difficult it is).

blog comments powered by Disqus