The proceedings of the 13th international ACM SIGACCESS conference on Computers and accessibility - ASSETS '11
Full text: Download
The use of visual aesthetics has been found to contribute to feelings of a positive Web experience. Accordingly, studies report strong correlations between aesthetics and facets of user experience like usability and credibility, but does this hold for accessibility also? Some believe that Web aesthetics impedes accessibility, while most Web designers perceive that the accessibility initiative is restrictive design-wise. These misconceptions have slowed down the advancement of an inclusive Web. Firstly, it is clear that the relationship between Web aesthetics and accessibility is still poorly understood. Secondly, tools capable of analysing the aesthetic quality of Web pages and relaying associated accessibility status information are lacking. This thesis addresses these two problems. In order to investigate this relationship, the aesthetic judgements of 180 users were elicited to help classify Web pages based on their visual quality using Lavie and Tractinsky's framework. The classified Web pages were then technically and manually audited for accessibility compliance using 4 automated tools, and 11 experts who used a heuristic evaluation technique known as the Barrier Walkthrough (BW) method to check for barriers which could affect people with visual impairments. Our results consistently showed that Web pages judged on Lavie and Tractinsky's classical aesthetic dimension as being `clean' had significant correlations with accessibility, suggesting `cleanness' to be a suitable proxy measure for accessibility. Expressive dimensions showed no such correlations. This insight was used to develop the EIVAA tool aimed at predicting the aesthetic quality of Web pages and using the information to provide accessibility ratings for the pages. Quantitative evaluations show that the tool is able to predict aesthetic quality in a way that mimics gold standards, especially along the design dimension `clean' where we observed tool-human correlations as strong as 0.703, thus making the associated accessibility predictions also acceptable. We envision that our findings will give the Web community a more holistic understanding of the interactions between the use of aesthetics and accessibility, and that our tool would inform Web developers of the implications of their designs.