Review: Analysing graphical readability

Already two weeks ago I cleaned my desk at home and sometimes it is useful to clean the desk seriously. In this case it was not the usual move-the-piles-around exercise but serious revisiting the information that has piled up. Getting through all the notes, business cards, bills and papers on my desk was almost archaeological work - after only three months of ignoring my IN tray ... well, at least I stumbled upon an article from Elzer, Green, Carberry and Hoffmann which has been published last year in UMUAI. The article has the wonderful title "a model of perceptual task effort for bar charts and its role in recognizing intention". I already read this article some time ago and I remembered that it was somewhat related to my research. However, that title of the article is already somewhat mind-puzzling and I forgot what it was all about and thus I read it again.

Somehow I had the feeling that this might be a bad idea, but I thought that it is useful even after the submission of my review article. It turned out that the article was lying on my desk long enough for me to forget about it and its relevance for the theoretical background of my research.

The authors analysed the ease of understanding different bar charts. Unfortunately, the authors focused too much on the readability of graphical representations and forgot about the perceptual task effort of reading text. However, even if text scanning and speed reading techniques fail with this article, it has its value that makes it worth working through the text, thoroughly.

The core idea of the article is that there are general rules that describe the ease of perceiving bar charts. These rules may help designers to organise the graphical information in a way that makes it easier for readers to understand the meaning of a graphic. The authors highlight that such perceptual task effort must not be confused with the cognitive task effort. While the prior addresses the process of extracting and organising information from a graphic, the latter describes the process of understanding the information. Therefore, I prefer graphical readability over the authors' perceptual task effort for describing the main concept of the article.

The authors use Bayesian rules to model the complexity of a bar chart with regard to graphical readability. These rules were validated experimentally with a small user group and by using user interaction monitoring and eye tracking techniques. Although the results of the experiment were neither surprising nor new, the rule based model provides some interesting implications. First, that reading a graphic requires effort that can be modelled by a fairly small set of rules. Second, that graphics provide implicit tasks to the users, such as finding maximal and minimal values. Third, following the rules for graphical readability help to underline the meaning of information.

For my research this is relevant for applying such rules for guiding the users' attention. We have to keep in mind the perceptual tasks and cognitive tasks are not entirely independent. Therefore, by carefully increasing the time a user has to "read" graphical information we can add some time for reflecting on this information. With regard to smart indicators we can hypothesize that task critical information is better recognised if the slightly more perceptual effort is required than for information that is non-critical to a task.

Full Reference

Elzer, S.; Green, N.; Carberry, S.; & Hoffmann, J. (2006) "A model of perceptual task effort for bar charts and its role in recognizing intention" In User Modeling and User-Adaptive Interaction (UMUAI), 16 (1), pp. 1-30. Online version