Adaptive interfaces : cognitive styles and personality characteristics as determinants of support.

The main purpose of the thesis was to examine the human computer interface from a user-centred viewpoint in order that it adapt to the characteristics of the user in terms of his/her personal style (cognitive and learning styles, personality traits). The relationship between personal style and performance, and comprehension of the system's operations are explored in order to highlight the characteristics that are most likely to predict the type of user that is the most successful learner. The manner in which the interface can support the user was explored in terms of the type of environment best suited to each individual, as was the use of colour (non-qualitative) to provide feedback as to the user's location within a system. Qualitative colour was also examined in relation to performance and the potential of using qualitative colour coding to enhance the user's response. Two methodologies were employed: applied (Studies 1-3) and naturalistic (Studies 4 and 5). The former attempts to simulate an environment which forces the subjects to use information processing strategies as if they were in a naturalistic environment. The findings of Studies 1 and 2 indicated that the Embedded Figures Test (EFT) is a good predictor of success in a simulated word processing environment. Field independent subjects were found to consistently perform better than their field dependent counterparts. In Study 1 it was expected that colour would act as a support to the user and would therefore benefit field dependent users. The results suggested that colour hindered performance but this may be due to overuse. Study 2 also investigated the usefulness of coding information. Colour and shape coding were found to be equally good as cues for remembering over no form of coding. Many of the studies which have investigated the use of colour to enhance performance have relied on redundant colour coding of arbitrary symbols. Study 3 explored the usefulness of qualitative colour coding and meaningful stimuli to determine the effects on performance and the potential of using qualitative colour coding at the interface. The results showed that there are no differences in response time for qualitative colour coding, non-qualitative colour coding and no colour coding. The lack of significant findings suggested that position of the stimuli may have confounded the results. In a naturalistic study (Study 4) it was possible to tentatively predict from the subject's personal style, the type of software environment most suited to him/her through observation of his/her behaviour in addition to the types of problems encountered by subjects and their frequency of occurrence. Study 5 attempted to relate natural language and comprehension to personal style. The findings suggested that question-asking techniques do not appear to be a contributor to successful learning. Additionally, the most efficient learner in terms of the number of tasks that s/he was able to complete and scores on a comprehension test, was the redundant holist and holist. Serialists were found to be slower than holists, and thus the design of the study may have contributed to these findings. Taken together, the studies suggest the type of user who tends to be most successful in interacting with computers in terms of behavioural data such as number of keystrokes, frequency of asking questions, frequency of asking for help, time spent reading the manual etc, in terms of the misunderstandings experienced and in terms of their comprehension of system operations. The use of colour as a support to less able users was not borne out in the study and may be due to the care that is required in employing colour at the interface. Colour as an associative cue maybe effective in comparison to no colour cue. However, other forms of cueing may be equally effective, such as shape coding.