Research Report—A Confirmatory Factor Analysis of the User Information Satisfaction Instrument

Abstract

The structure and dimensionality of the user information satisfaction (UIS) construct is an important theoretical issue that has received considerable attention. Building upon the work of Bailey and Pearson (Bailey, J. E., S. W. Pearson. 1983. Development of a tool for measuring and analyzing computer user satisfaction. Management Sci. 29(5, May) 530–545.), Ives et al. (Ives, B., M. Olson, J. J. Baroudi. 1983. The measure of user information satisfaction. Comm. ACM 26(10, October) 785–793.) conduct an exploratory factor analysis and recommend a 13-item instrument (two indicators per item) for measuring user information satisfaction. Ives et al. also contend that UIS is comprised of three component measures (information product, EDP staff and services, and user knowledge or involvement). In a replication using exploratory techniques, Baroudi and Orlikowski (Baroudi, J. J., W. J. Orlikowski. 1988. A short-form measure of user information satisfaction: A psychometric evaluation and notes on use. J. Management Inform. Systems 4(4, Spring) 44–59.) confirm the three factor structure and support the diagnostic utility of the three factor model. Other researchers have suggested a need for caution in using the UIS instrument as a single measure of user satisfaction; they contend that the instrument's three components measure quite different dimensions whose antecedents and consequences should be studied separately. The acceptance of UIS as a standardized instrument requires confirmation that it explains and measures the user information satisfaction construct and its components. Based on a sample of 224 respondents, this research uses confirmatory factor analysis (LISREL) to test alternative models of underlying factor structure and assess the reliability and validity of factors and items. The results provide support for a revised UIS model with four first-order factors and one second-order (higher-order) factor. To cross-validate these results, the authors reexamine two data sets, including the original Baroudi and Orlikowski data, to assess the revised UIS model. The results show that the revised model provides better model-data fit in all three data sets. Thus, the evidence supports the use of: (1) the 13-item instrument as a measure of an overall UIS; and (2) four component factors for explaining the UIS construct.