A Discrimination of Software Implementation Success Criteria

Software implementation projects struggle with the delicate balance of low cost, on-time delivery and quality. The methodologies and processes used to create and maintain a quality software system are expensive to deploy and result in long development cycle-time. However, without their deployment into the software implementation life-cycle, a software system will be undependable, unsuccessful. The purpose of this research is to identify a succinct set of software implementation success criteria and assess the key independent constructs, activities, carried out to ensure a successful implementation project. The research will assess the success of a software implementation project as the dependent construct of interest and use the software process model (methodology) as the independent construct. This field research involved three phases: (1) criteria development, (2) data collection, and (3) testing of hypotheses and discriminant analysis. The first phase resulted in the development of the measurement instruments for the independent and dependent constructs. The measurement instrument for the independent construct was representative of the criteria from highly regarded software implementation process models and methodologies, e.g., ISO9000, Software Engineering Institute's Capability Maturity Model (SEI CMM). The dependent construct was developed from the categories and criteria from the Delone and McLean (1992) MIS List of Success Measures. The data collection and assessment phase employed a field survey research strategy to 80 companies involved in internal software implementation. Both successful and unsuccessful software implementation projects (identified by the Delone/McLean model) participated. Results from 165 projects were collected, 28 unsuccessful and 137 successful. The third phase used ANOVA to test the first 11 hypotheses and employed discriminant analysis for the 12th hypothesis to identify the "best set" of variables, criteria, that discriminate between successful and unsuccessful software implementation projects. Twelve discriminating variables out of 67 were identified and supported as significant discriminators between successful and unsuccessful projects. Three of the 11 constructs were found not to be significant investments for the successful projects