Friday, October 5, 2012

ready, set, critique!

I just came across the following daily chart from The Economist on my Google+ stream:

See full article.
My questions to you: What story does it tell? What story could it tell? How would you change the way the information is presented to do that in an effective way?

Ready...set...critique! Leave a comment with your thoughts.


  1. The chart says more women are winning the fellowships on a percentage basis and the average age has been relatively consistent over time.

    Ignoring for the moment that the criteria for this award is secret and therefore probably can't generate meaningful data that would "help" a potential award winner, what I'd want to see the chart show me is whether the average age of female winners is different from the average age of men.

    After reading the story, I realized that the average line is misleading because of the range of ages (18-82) of the winner.

    Given that the story is essentially a data dump without linking us to the data, I guess I'd like the chart to focus more on the ages by gender and the line to be the percentage of male or female winners. While I'd rather get a sense of the range (are there more younger winners than old but really old people are skewing the data), I guess the challenge is that your denominator is so low that it would be difficult to avoid sweeping generalizations.

    Additionally, this approach would enable me to create lines with additional percentage data for different diversity categories (e.g., race, geography)

  2. I really don't like this one. a few things jump out at me the 1st of which is the axis's use a similar scale but are different measures (count vs age), also what's the value of average age? I want to know the distribution and even better would be to know the distribution by gender as well. It would be interesting to know if there's a difference and is that consistent over time or has it been trending? I'm sure the data is available and would not be overly complicated to display. :-D

  3. Not a big fan of the inconsistency with the x axis...regardless of whether or not there are 1, 4, or 5 years between bars, they should be spaced out evenly. This is particularly true when using line charts to not misrepresent the rate at which the data is increasing or decreasing.

  4. Well, One thing I immediately dislike are the stacked bars, two axis on the same chart, where the author did not even take the time to put a proper scale on. Also using the absolute numbers makes it difficult to see whether the proportion of male vs. female did actually change a lot (apart from 1995). A better idea would be to show simply the percentage of female, and using something advanced, like parallel boxplots for the age distribution, of course in two graphs... This would tell the whole story in two neat displays.

  5. Here are my comments on this confusing visual. (1) If you don't actually read the values on the X axis, you wouldn't realize that the first interval is 4 years, the next four intervals equal 5 years each, then it goes by individual year after that. That's part of why the horizontal line for "average age" looks so much flatter on the right side of the graph. Also, why is the line maroon but the label at the top for "MacArthur Genius Awards" is a different shade of red? (2) The use of the two shades of blue for the stacked bars did a pretty good job of representing the proportion of each gender, i.e. earlier on it was mostly males and more recently it's closer to evenly split. (3) The two values for the Y axis had me baffled. The one on the right is labeled "average age", and I assume that is supposed to correspond to the maroon line that runs across the graph above the bars. But what are the numbers on the left side of the Y axis supposed to represent? Number of award winners? If so, then it seems that the message is that they give out fewer awards now than they did earlier. I briefly considered whether it meant the number of FEMALE award winners (because the blue used for those labels is the same color blue as the female portion of the stacked bars). Suggestions for improvement would be (1) Use even increments on the X axis. One year, two years, five years, whatever -- they need to increase evenly. (2) Don't use the same shade of blue for the female portion of the bar as used on the left side of the Y axis, unless it's supposed to mean something. (3) Using median age would be better than mean age if the range is so great. I still don't quite understand the use of the Y axis for two things. I think they should either choose one message (i.e. age or count) or make another graph. It's too much information for one graphic, IMO. Nicole