An investigation of object-oriented interfaces for human computer interaction

The thesis reports an investigation of object-oriented interfaces which was undertaken to advance understanding of their implications for the user. The object-oriented interface is a comparatively recent development in user computer interface technology but, despite its popularity and reputation for ease of use, little research about it has been reported from a cognitive ergonomics perspective. The investigation involves two activities, one analytic and one empirical, aimed at providing a characterization and assessment of object-oriented interfaces. The analytic activity specifies a framework within which object-oriented interfaces can be described and the empirical activity yields experimental findings concerning the usability of object-oriented interfaces. First, the framework is proposed. It comprises some distinctions, definitions and a levels structure for modelling devices and users. Throughout the investigation, the structure is applied to derive particular models of devices and user behaviour. This process enables elements of the framework to be validated or modified. On the empirical side, an observational study is conducted which compares user performance on an object-oriented interface and a non-object-oriented interface. It determines that object-oriented interfaces are easier to use and that the dimensions for classifying object-oriented interfaces have consequences for user performance. Following the observational study, a specific device feature, syntax of the command sequence, is selected for further experimentation and a rudimentary user model is proposed. Two experiments, using simulated interfaces, are conducted to discover the effects of I/O Level and Task Level factors on syntax usability. The results of these studies advance the user model. These studies suggest there is no inherent difference between the two command sequences tested. An alternative view is proposed which suggests that performance differences found in the observational study are due to characteristics particular to objects which were not tapped in the experimental studies. A third experiment is performed which lends some support this proposition. The thesis concludes that: i) an object-oriented interface can be distinguished from other interface classes in terms of features which have consequences for the user; ii) under certain conditions, object-oriented interfaces are easier to use than their contrast and the object-oriented variant of the feature, syntax, has a facilitatory influence on user performance