Designing for older users is designing for all. A teleshop system was designed for all ages. Existing teleshop systems were evaluated in three experiments with users of different age and with different levels of computer experience. Subjects had to perform the same shopping task in all experiments: ordering a list of groceries and making some changes afterwards. Measured was the time to perform the task, the number of mistakes and the amount of help from the experimenter. Older users had more problems than younger users and needed more help. Users with more computer experience performed better. All teleshop systems were analyzed using the GOMS (Goal, Operator, Method, Selection Rules) method of task analysis. A more detailed analysis of the task gives information which bottlenecks for which users exist. For example: older users need more time than younger users to learn a new task. The task analysis shows that typical computer language is frequently used. This is difficult for inexperienced users. Furthermore, the system is inconsistent, i.e. different methods and operators have to be used for the same sub-task. The first problem is reduced by using common world knowledge. The second problem by designing an interface with more consistency.

A new teleshop system was designed based on the outcomes of the three experiments and the task analysis of the systems. The new interface applied to existing knowledge of real shopping. Normal guidelines for designing interfaces were followed. The teleshop system runs on a normal PC with a mouse as input device. To start the shopping task users have to walk through the door of the mall using the mouse and thereby using their mental model of real shopping. To evaluate the new system it was compared to an existing CD-I teleshop system with the same structure. Three groups of users, ranging from 40 to 80 years had to perform the standard task on one of the systems. Each group was matched on computer experience and level of intelligence. The results showed that the new system was very user-friendly and all groups performed more or less well. The existing system gave more trouble to subjects with less computer experience and a lower level of intelligence. Some failures remained however in the new system. Changing the groceries list was a new task and this caused trouble. When subjects were reminded of their mental model of real-life shopping they could solve the problem. On the other hand mistakes were made where the system did not completely match the real world. In order to create a more realistic system, users where shown graphical representations of the products. The researcher had the possibility to choose between a graphical or a textual representation.

Within the new system a component was built to register al the mouse movements and mouse clicks. With this information and a new version of the system, which was specially developed for this purpose, it was possible to replay an entire session on the PC. With the combination of both the replay program and a program that registers all events (including the mouse movements and mouse clicks) the researcher could easily evaluate specific problems in task performance.

A demonstration of the new system will be given at the conference, including a demonstration of the event replay program.