Human error contributors in computer interaction: Analysis of expert and novice performance in complex task accomplishment
Jeffery V. Mosley, Wayne State University, United States
Wayne State University . Awarded
The impact of interface design on operator performance is not an exact science at this time. There is no theory base available, only guidelines and best practices. Further complications may be inferred that the operator environment, personal state, and the state of others may affect individual and group performances in complex task execution.
The thought that an individual can have an emotional reaction to the interface, whether positive or negative, is important as the reaction can directly influence task performance. Efforts to train individuals and groups within environment using interfaces that are too complex or confusing also can contribute to the level of task performance.
Relevance is meant to mean the level of contribution of stress to human performance and the level of the correlation of the relationship between stress and performance.
This study attempts to link the occurrence of error to degraded performance, specifically through review of research previously conducted by the Army Research Institute (ARI) to identify performance differences between novice and experts groups during the execution of a complex task.
This was an ex-post facto analysis of self-reported data on performance and individual perceptions defined in the experiment. The research attempted to statistically correlate performance to reported levels of stress and frustration for novice and expert operators from the ARI experiments.
Data from the original experiment were analyzed using Pearson r correlations of ARI experiment participant self-reports of stress related to self-reported operator performance. HCI error manifestations were assessed using frequency of occurrence data from the original experiments.
The procedure for validating relevance of the affective contributors from the human computer interface and interaction were based ARI experiment task analysis and correlation analysis (i.e. learner performance to self-assessment of the levels of stress and anxiety during the experiment). The objective was to define the contributors, quantify the correlations of the contributors to human performance in the training and natural work environment for the complex task.
The results of this study have been presented to present the analysis of participant data both subjective and objective from the ARI experiments. The analysis attempted to relate performance to error and error to situations of high stress. The ARI data alone does not fully show correlation to support the antecedent relationships for error, stress, and human performance. The study does provide inference that certain frequency of human computer interaction; levels and types of participant verbalizations can potentially create an error situation in the execution of a complex task. The inference is supported by analysis of ARI data comparing novice and expert interactions within the unit cell. The study does identify that the novice frequency of interactions were less than those of the expert for the same tasks.
The study provided a narrative review and comparison of human reliability models and techniques that are currently used in industry. The models were mainly based on task analysis and subjective probabilities of error for a particular task and not viable to identify the affects of stress to create error, or its impact on performance.
Mosley, J.V. Human error contributors in computer interaction: Analysis of expert and novice performance in complex task accomplishment. Ph.D. thesis, Wayne State University.
Citation reproduced with permission of ProQuest LLC.
For copies of dissertations and theses: (800) 521-0600/(734) 761-4700 or https://dissexpress.umi.com