In Antibiotic Effectiveness – A Study of Chart Types I followed a path through a number of graphical approaches to analyze data on effectiveness of antibiotics first tabulated by Will Burtin in the 1950s. I entered my final visualization in a contest sponsored by Chance Magazine.
Xan Gregg shares his own analysis of the antibiotic data in Burtin Antibiotic Illustrations. I recommend that you read his post to get another look into the thought processes behind such an analysis. Xan works for JMP, maker of statistics discovery software, which is in turn a division of SAS, which produces business analytics software. Small world. I first learned of JMP software when I was looking for a name for my company: I Googled for my initials (“JMP”) and found them already in use for graphics and analytics software.
Xan’s analysis started with questioning what was wrong with the data table. Nothing really, except any patterns are hidden by the alphanumeric characters, and as I pointed out, sorting alphabetically by bateria name further obscures any patterns. Xan applied a heat map to the tabulated data as a first look at any extreme values or patterns.
Xan then used a dendrogram in conjunction with his heat map to identify similarities among pairs and groups of bacteria. Such groupings could help identify areas of investigation.
Since there were three antibiotics, Xan made a three dimensional view of their effectiveness, and posted two views, which indicate there may be two groupings of behavior. Again, areas for further investigation.
Xan flattened this 3D view into a scatterplot matrix, showing all three of the two-dimensional charts.
Finally, Xan thought of the 1950s physicians who may have no idea what’s making his patients sick, but wants an effective treatment. The following panel chart shows an interesting and elegant observation: Penicillin is most effective against Gram-positive bacteria, Neomycin is most effective against Gram-negative bacteria, and Streptomycin is the best choice when Gram-staining response is unknown.
When I undertook my analysis, I spent half an hour Googling for Burtin’s original graphical analysis, but I came up empty. Xan was more determined than I, and found the graphic below in a NY Times book review for Design and Science: The Life and Work of Will Burtin.
Well, that’s pretty bad. About the only thing he did right was to make the bars indicating lower antibiotic concentrations (greater effectiveness) longer. Xan has this to say about Burtin’s graph:
It has little communication value except to say “Look how cool I am!” At least all the data is present, so a meticulous reader can get what information he needs. The audience for this must be a hospital administrator who needs to feel like he’s getting his money’s worth with fancy visualizations. I think it is more a work of art than of communication.
Very well said.