Monday, September 12, 2011

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

The paper written in 1984 is an attempt to make a basic theory of graphical perception. The theory they set out to specify and validate is not exhaustive but still sets the course for the future developments. Based on experiments and the theory of Psychophysics they formulate an ordering of perceptual tasks. The ordering lists the perceptual tasks in the increasing order of the accuracy of perception by humans. This divides the 10 visual perception tasks taken into consideration into 6 groups. Where each group has one or more than one perceptual task and has lower perception accuracy when compared to the group before it in the list. Though the groups seem to work for the experiments done by the authors. There is not enough explanation as to why some of these perceptual tasks were considered to have the same accuracy. There was no experimental evidence given to prove that perception of length, angle and direction can have the same level of accuracy. There is sufficient discussion and reference to experiments to prove the ordering. There could have been some discussion on how the order is affected when the amount of information being represented increases.

Based on the experiments and theory described in the first part of the paper the authors have shown a new kind of charts which are evidently better compared to the traditional representations. It is very convincing to see that the grouping dot chart can convey information more quickly and easily when compared to grouped bar chart. The framed rectangle chart when compared to the shaded area chart is better to compare any two states and their murder rates visually. However in my opinion the shaded area chart is better in representing parts of the country where there is high murder rate and where there is lesser murder rate. The shading dominates the state boundaries and hence provides a more unified view. 

This theory in general forms a basic idea based on which visualizations and choices made to represent data can be justified. However a lot of work needs to be done to make this a more exhaustive and universal theory for data visualization.