When Cutesy Is Confusing in Graphs

A colleague recently sent me a link to a Stubborn Mule blog post titled “Pyramid Perversion – More Junk Charts.” The author refers to a graphic (shown below) that came from the Consumerist. The Stubborn Mule author had a number of problems with it. I have even more.

The graphic tried to suggest that federal subsidies for food production don’t match up to the importance of each food group. The creators chose the food pyramid for their visual. While this may have seemed like a clever mix of the recommended serving graph and subsidy data, it fails to properly display the numbers. It’s more cutesy than practical and makes it hard for the viewer to make any real comparison. You would need to match the food type (by colored sections with labels) to see the apparent discrepancy between the data. It’s also a clear violation of Edward Tufte’s rule of graphs: “Above all else, show the data.” Not only is a pyramid a poor choice of graph type, but the scales are also different for each one, i.e., percent of subsidies vs. number of servings.

A better way would have been to first convert the recommended servings by food type into percents (by taking the number of serving for each food type divided by the total number of servings), and then use a bar graph as follows:

The bar graph clearly shows the difference between what’s deemed important to a healthy, balanced diet vs. what’s subsidized. The viewer doesn’t have to shift his or her focus from one pyramid to the other. The data has been normalized to allow the author to plot the data on one scale and have one visual to show the results.

Are there any other suggestions on how this data could have been presented better?

tags: Data Visualization, JMP 8

2 Comments

  1. Dave Garbutt
    Posted April 4, 2010 at 7:50 pm | Permalink

    Great example and interesting question.
    I think it would display better as a spine plot (also with other names) eg butterfly plot.
    The point of he chart is , I think, that the subsidies are not going to the most recommended foods.
    So two back to back histograms will show this very well.
    Not sure if this is possible in JMP but it can certainly be done in SAS 9.2.
    (I will have a go when I get back to the office & my copy of SAS)
    Dave

  2. Dave Garbutt
    Posted April 4, 2010 at 7:57 pm | Permalink

    OOPs I for got to mention:
    Getting the sort order right makes a big difference, I think the categories should be sorted on portions/serving so the largest number of servings is at the bottom of the chart.
    ie (Top down)
    Sugar, Alc,
    Proteins
    Veg+ Fruit
    Grains.
    This then echoes the pyramid theme - at least on the servings side.
    Dave

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <p> <pre lang="" line="" escaped=""> <q cite=""> <strike> <strong>