Application software documentation : a reader measure
The current trend of end-user computing in organizations implies the increased exposure of people with differing backgrounds and skills to the use of management information systems (MIS). Documentation constitutes a significant, primary interface between the individual, whose job tasks are impacted by computer-based applications, and the application system or software package. The professional literature has repeatedly noted the lack of adequate, meaningful documentation in practice. The MIS literature has identified documentation as a factor influencing user satisfaction, project success, software maintenance, and application package usage. The MIS literature has generally supported a relationship between the use of application software and the beliefs and attitudes of its users. The underlying perspective of this study regards information about application software provided through documentation as a psychological product. Documentation meaningfulness is defined in terms of the aggregate judgmental responses of the reader to the documentation. An extensive interdisciplinary literature review, field interviews, and an expert review were conducted to establish theoretical and empirical bases for scale construction and selection. The study questionnaire, including 70 semantic differential scales, was distributed to over 30 primarily Houston-based organizations, and 483 usable forms were returned. Each respondent was asked to evaluate documentation most familiar to him in the performance of job-related software tasks. Upon receipt forms were alternately assigned to validation and cross-validation groups. Exploratory factor analyses on validation data yielded a derived measure consisting of three principal factors: task relevance, systematic arrangement, and credibility. Confirmatory factor analysis using the factor model derived from the validation group analysis was conducted, identifying the factor pattern on cross-validation data. Factor stability was supported by high indexes of factor similarity for solutions of validation and cross- validation data. Satisfactory reliability estimates (coefficient alpha) were obtained for the factors and the derived measure.