| This dissertation represents exploratory research intended to validate an entropy model for information systems management that suggests that an information system's problem reporting data can be used to calculate an objective measure of entropy using Shannon's entropy equation from the domain of information theory. The resulting measure of entropy, an indicator of system disorder, could be used to aid information system managers in various decision-making processes. To test the model, entropy measures were calculated from the problem reporting data of the Global Command and Control System (GCCS), a federation of approximately 40 different command and control applications, and compared to the results of an end user satisfaction survey conducted on the same system. Both application-level and system-level relationships were investigated. At the application level, the results indicated a medium negative correlation between entropy and end users' perceptions of satisfaction and supported the most general hypothesis that there is indeed a relationship between the two variables. The results did not support a more specific hypothesis that the entropy associated with “High Priority” problems was more of an indicator of user satisfaction than “Low Priority” problems. Actually, the results suggest that the converse is true. In the course of testing this hypothesis, the results also suggest that it may not be advantageous to sub-categorize problem reports for the purpose of calculating entropy. At the system level, the results indicated that, when using entropy as an indicator of system-level satisfaction, the entropy associated with just the general, non-application-specific problem reports is the most accurate indicator of system-level satisfaction. Using measures of system-level entropy based on the total number of application-specific problems was not as accurate an indicator of user satisfaction as was the measure based on general system-level problems. Finally, an original assumption was that the entropy model would prove to be an objective tool to use in information management. However, it was discovered that a small degree of subjectivity cannot be avoided when classifying problems and analyzing problem types for uniqueness. |
