Statistician Heal Thyself: Have We Lost the Plot?

Gordon, I., & Finch, S. (2015). Statistician Heal Thyself: Have We Lost the Plot? Journal of Computational and Graphical Statistics: A Joint Publication of American Statistical Association, Institute of Mathematical Statistics, Interface Foundation of North America, 24(4), 1210–1229.

  • this is an article about graphics in science and statistics journals
  • I think graphical display of data is much easier to parse than tabled data

    • but many (applied) journals specify a maximum of figures per article!

    a high quality graph should be able to be interpreted without elaboration

  • Figure 1 shows a cluttered display

    • I think there\’s a ggplot2 function to push labels apart
    • yes, here it is: ggrepel
  • Figure 2 provides a redesign

    • basically a small multiples plot
    • this is an improvement, but if the relationship is not monotone, then this figure would look a bit messy
  • Figure 3 provides a scatterplot alternative

    • the authors suggest that it is easy to see that average lifespan increases with education, but I find this difficult
    • I can more easily see the decrease in lifespan variation with increasing education though
  • Principle 3:

    Use good alignment on a common scale for quantities to be compared.

    • pie charts and stacked bar charts fail this principle
      • stacked bar charts because only one category (the bottom one) is aligned against the common axis
    • panel plots also fail this?
  • The figures in the article themselves are fuzzy

    • perhaps the authors could have taken their own advice?!
  • Detection was found to be the main issue

    • This refers to how we find out about features of the graph
  • They provide a checklist at the end of the article for those producing graphics