This report is also available as an Acrobat file.

Contents
Authoring and Design for the WWW

DESIGN INTO PRODUCTION

User-centred design

The design process

Nielsen gives an account (in User Interface Design for Sun's WWW Site', see Web Resources) of the design process for a Web site. Different publishers and designers will adopt different approaches, but it may be useful to outline how Nielsen's team set about the task, especially in their use of off-screen' methods.

Initial design was done away from the computer as a card-sorting test. A number of functions and areas which the site was intended to support were written on individual cards and a small group of subjects was asked to group the cards into logical sets and apply a hierarchical order within the set. This is more efficient than beginning by coding several HTML pages. It can also allow early elimination of some problems, for example terms which no subjects understand, and omissions and inclusions which subjects object to.

In the Sun site a prior decision had been taken to use icons, so another component of their design process was the trialling and modification of icon designs both individually and as a set. Subjects were asked to match the icons to the previously devised categories from the card-sorting exercise.

Main screens were mocked up on paper and subjects were asked to point to the various components of the page and say what they thought they did and which other elements they led to.

It should be noted that the trialling of ideas for the site was reactive. The system designers made proposals which were then modified in the light of feedback. This is generally preferable to asking users what they want, since this can lead not as might be thought to too many ideas, but to a paucity of ideas where all those questioned answer only on the basis of other, perhaps inadequate, systems which they have seen. Nevertheless, the ab initio approach has its place, especially where a constructive brainstorming session with intended users is possible.

Usability testing

Without testing, user-centred design is a sham. It is essential to find out what users really do when they encounter the site. There are many ways of testing information systems, from quantitative methods tested with scientific rigour through to informal observation. Publishers will choose the methods most appropriate to their site and their needs, since clearly there is a difference in the levels of complexity of pages and sites, and even more importantly between the purposes which sites are designed to serve.

It has been pointed out that there is no such thing as the average user. It is important that testing takes place using subjects who represent the actual intended audience. For informal testing aimed at improving a site rather than at proving a methodological point, it is also preferable to test in realistic circumstances. So, for example, if an on-line course module is developed as part of a suite of activities, it should properly be tested in a simulation of the real usage, rather than in isolation. Of course in some circumstances these are counsels of perfection. The key point is that any testing is better than no testing.

Aspects which will require testing include technical tests in various browsers, under different operating systems and on varying hardware, both locally and remotely. However these technical tests are only part of the story user-testing of the content, structure and interaction is just as important.

Some test methods

Test methods (some of which are likely to beyond the budget of most HE developers) include the following

Direct observation of users: watching the system in use and making notes
Indirect observation, for example using video recording
Think-aloud, where users describe what they are doing or trying to do
Software logging: a number of server-based programs are now able to monitor what users are doing (though not what they are thinking!)
Post-hoc interviews and questionnaires A readable account of the strengths and weakness of these methods is provided in Preece 1994.

Some test findings

Sun's usability studies suggested to them that:

People have little patience with poorly designed Web sites.
Trust of the content in a site is based partly on a perception of its organisation and design.
Users do not like scrolling. Information not at the top of the screen is only read by very interested users.
Reading speeds are 25% slower on screen than from paper. To compensate, users recklessly skip over any text that they deem to be fluff, and scan for highlighted items.

Contents

Graphics Multimedia Virtual Environments Visualisation Contents