Abstract

To support the human factors engineer in designing a good interactive system, a method has been developed to analyze the empirical data of the interactive decision behavior described in a finite discrete state space. The sequences of decisions and actions produced by users contain much information about their mental models, the individual problem solution strategy for a given task and the underlying decision structure. We distinguish between (1) the logical structure (the 'device model'), (2) the sequential goal structure, and (3) the temporal structure. The analyzing tool AMME can handle the recorded decision and action sequences and automatically extracts a net description of the task dependent decision model (the logical structure). This basic model is extended by further elements to reconstruct an empirical expert user sequence. This article presents two modeling strategies: (1) event-driven, versus (2) parallel goal setting processes. Both strategies add sequential structure to the logical structure. Three different models are presented and their predictive power is discussed.

Behaviour of expert and novice database users solving the same task was recorded. Several successful strategies were identified. Since there are more users than strategies, some users applied the same strategy. The aim was to develop methods grouping users with common strategy. Following three approaches (correlation, intersection, and exclusion), a metric among task solving behavioural sequences was defined. Measured data was organised in matrix systems relating all users. Statistical and analytical interpretation of matrices showed distinct groups. A common denominator for a group can indicate a strategy.

No abstract

No abstract