Wednesday, September 7, 2011

Reaction: Graphical Perception: Theory, Experimentation and the Application to the Development of Graphical Models

I appreciate that the authors set out by saying that there had been no prior scientific discussion of graphs, just a smattering of effectiveness comparisons, and some subjective comparisons. Their mathematical calculations really drive home their point (e.g. log2( 1 judged percent - true percent I + 1/).

I do not, however, agree with their seeming conclusion that dot charts can be used basically everywhere. Their example for replacing a pie chart with a bar chart, and then a dot chart, seems absurd on its face. They ask you to order the pie slices visually which is hard, but I think if you put the slices in order in the first place it's trivial. Plus, you can label each pie slice easily. From the dot chart it's not all clear that the percentages add up to 100%, and it also doesn't fit with any paradigm that I've ever seen. I think we're conditioned to expect pie charts when measuring percentages. I'm fine with their conclusions about angular, length and position differences, but it seems mis-applied here.

Completely off-topic, but the paper loses points for putting the figures/images that are discussed two whole pages after the discussion about them, rather than before or immediately afterwards.