Accessibility for All and tacit knowledge requirements of ICT systems

Gill Whitney is a charted engineer and a Member of Institution of Electrical Engineers (MIEE). She has a M.Sc. in Information Engineering - Specialisation in Measurement and Information in Medicine from City University and a B.Sc. (Hons.) in Electronic Engineering from Kent University Canterbury. She is a Senior Lecturer in the School of Computing Science at Middlesex University. She is Chair of the European Design-for-all and Assistive Technology Standardisation Co-ordination Group (DATSCG). http://www.cenorm.be/isss/dfa-ict/ and she is a Member of the Design for All Working Group of ANEC (ANEC is the European Association for the co-ordination of consumer representation in standardisation).
http://www.anec.org/

Ray Adams is a qualified, PhD, applied cognitive psychologist trained in London and Churchill College, Cambridge. He has active research and teaching interests in universal access, inclusive design, assistive technology, the psychology of computing and research methods. He is a Senior Lecturer in the School of Computing, Middlesex University, London.

Familiarity is increasingly seen as an important factor in the effective use of information technology and website design (Lohse, 2000). These findings imply that users must hold a considerable amount of information in their heads to use websites successfully. It is likely that designers, who are familiar with their designs, may underestimate the levels of tacit knowledge that their websites require of the people who use them. People who have a sensory impairment have an additional problem when using these websites as they already need to have a high level of knowledge to use their access technology, in addition to the knowledge required to use the websites. This work investigates how the levels of tacit knowledge required impair memory and comprehension. It has lead to the identification of design rules that can be used in order to assist designers to design accessibly.

There is growing research evidence for the importance of systems familiarity in system use. For example, Lohse (2000) reports studies in which users are found to adhere to familiar, known websites rather than to shop around for the best deals if that means using unfamiliar web pages. This is distinct from good system design for ease of use. A familiar but poorly designed system may be easier to use than a well designed but very unfamiliar system. We call this the familiarity paradox. Unfamiliar systems place a burden of learning on the new user. This burden may reduce accessibility for all and interact with an individual's disabilities. This problem is made worse when new websites demand that users must acquire unnecessarily high of information. If so, there is a need to identify how to balance the design and training implications of new systems which aspire to universal access.

These issues would not be important if all the systems experienced at work and travel were all highly familiar, well designed and simple to use. However, the familiarity paradox may also mislead experts in design since, by definition, they will be highly familiar with the systems that they are developing. There is a significant risk that they will consider their own systems easier than they really are.

Unfortunately, many systems in common usage have complex structures that can hide essential facts and require users to retain large amounts of information in long term or working memory. This is often because designers are mislead by the familiarity paradox.

There is an additional problem for disabled users, as they may become familiar with a piece of equipment that becomes unusable as they acquire a disability. If they have to transfer to using a different piece of equipment their ability to become familiar and confident with it may be hampered by their need to learn both how to use the piece of equipment and how to use it as a disabled person. This may involve the use of a piece of assistive technology or by using a different sense to obtain the output. In an ideal world all equipment would be designed so that the user would be able to continue using familiar equipment even if they acquired a disability.

To overcome the familiarity paradox problem, guidelines are required for designers to identify user knowledge problems. The aim is to identify the problems that a design for users unfamiliar with the system and with specific disabilities may face.

Some simple techniques are demonstrated to measure the levels of tacit knowledge required by systems in everyday use and how those system demands interact with disabilities of users. Guidelines for accessibility for all are presented which deal with knowledge demands and training requirements in the context of a simple working model of user knowledge.

There are a number of ways with which to evaluate the levels of knowledge required by different system interface designs. Designers may acquire some of these methods, or call upon fellow experts to assist with them. Thimbleby (2000) use the cost of knowledge graph to analyse the usability of a system based upon the system specification. Such a graph demonstrates the number of goal states which a user can gain access to against the number of action that the user must make to do so, i.e. the cost of knowledge acquisition for each state. The value of this approach lies in the opportunity it provides the designer to compare and contrast different website designs on the basis of the knowledge demands. The difficulty of this approach lies in the level of mathematical expertise required to implement it. It's strength is found in the level of objectivity and precision of expression that can be achieved.

A more qualitative approach is reported by Kitajima, Blackmon and Polson (2000), who have produced a series of models based upon Kintsch's (1998) well known construction / integration theory of text comprehension. The name of the CoLiDeS Model is an acronym for Comprehension - based Linked model of Deliberate Search. A central axiom of this approach is that the person viewing a web page must depend critically upon their knowledge in long term memory to understand the objects on a web page and how to navigate around the website. The user must use both top down (knowledge based) processes and bottom up (perceptual) processes in order to navigate a web-site.

The authors comment that a typical single web page may contain up to 200 objects that the user must identify and navigate. This clearly exceeds the limits of a person's short term memory (STM) and thus underlines the importance of long term memory (LTM) to hold considerable quantities of information. Perhaps designers should limit themselves to an average of seven key features per page if STM is to be of significant use by users. A key feature of this model is the emphasis upon the forward search through a web page. It is essential that when users selects the most obvious link, it should take them unambiguously forward in meeting their objectives. They call for the use of links that are clear and unambiguous in where they lead. They also point to the problems of using deep hierarchies in websites, which provide only very general guidance and a poor sense of direction for the user. This approach to website design is founded on the importance of comprehension text and graphics. Texts, links and images that are difficult to understand will produce websites that are difficult to use.

Kitajima, Blackmon and Polson (2000)'s theory predicts that deep hierarchies generate links which are difficult to comprehend, as compared with those of shallow hierarchies. Supporting evidence has been provided by Zaphiris, Kurniawan and Ellis (2002), who reported that users (both old and young) preferred shallow to deep hierarchies and also found the later to take more time use than the former.

When using this research with users with disabilities it must be remembered that there may be an additional layer of complexity between the disabled user and the website they are navigating. The access technology used may not be straightforward and it’s operation may not mesh with the operation of the application that it is being used with. For a user with a sensory impairment who is not using a piece of assistive technology additional problems may exist in that the user will be attempting to use the application in a situation where they are receiving less information than the designer anticipated. This will put more strain on the user’s memory.

Typically, information searches on the Web have disappointingly poor success rates at the present. The research considered here suggests that these rates may be due, in part, to the high demands that many, web page designers inadvertently place upon users, particularly upon those with visual impairments. When web links are difficult to understand, the above memory difficulties will be exaggerated (Kitajima, Blackmon and Polson (2000). If this scenario is correct, then some design guidelines may be identified.

The number of key objects per web page should be limited to a level that does not strain short term memory (STM) limits i.e. seven plus or minus two.

LTM is also of critical importance. Websites design should facilitate as far as possible the acquisition of relevant, knowledge in the long-term memory of users.

Since, texts, links and images which are difficult to understand will produce websites which are difficult to use, designers should give a high priority to the comprehensibility of the links in their websites, so that users can search more effectively.

This emphasis upon memory and comprehension as important considerations in web page design, also vindicates the design principles identified by Leporini and Paterno (2002). They investigated the potential relevance of a number of design criteria for web design for usability and accessibility for vision impaired users. Their criteria included; the logical partition of information, clear text for links, consistency of terminology and layout, limitations in the number of links and frames and the designation of importance levels to web page elements. All of these criteria are consistent with our current concerns with user memory and comprehension and the contingent difficulties for people with visual impairments who wish to use these websites.

Kintsch, W. (1998). Comprehension: A Paradigm for Cognition. Cambridge : Cambridge University Press.

Kitajima, M., Blackmon, M. H. and Polson, P. G. (2000). A comprehension - based model of web navigation and its application to web usability analysis. In McDonald, S., Waern, Y. and Cockton, G. (eds.). People and Computers XIV - usability or else. Proceedings of HCI 2000. London : Springer - Verlag. pp 357 - 374.

Leporini, B. and Paterno, F. (2002). Criteria for usability of accessible websites. In Carbonell, N. and Stephanidis, C. User interfaces for all. Proceedings of the 7th ERCIM Workshop, Paris pp 113 - 125.

Lohse, G. (2000). Usability and profits in the digital economy. In McDonald, S., Waern, Y. and Cockton, G. (eds.). People and Computers XIV - usability or else. Proceedings of HCI 2000. London : Springer - Verlag. pp 3 - 16.

Thimbleby, H. (2000). Analysis and simulation of user interfaces. In McDonald, S., Waern, Y. and Cockton, G. (eds.). People and Computers XIV - usability or else. Proceedings of HCI 2000. London : Springer - Verlag. pp 221 - 238..

Zaphiris, P. Kurniawan, S. and Ellis, R. (2002). Age related differences and the depth - breadth tradeoff in hierarchical online information systems. In Carbonell, N. and Stephanidis, C. User interfaces for all. Proceedings of the 7th ERCIM Workshop, Paris pp 453 - 472.