Eliciting User Feedback During System Development

Eliciting user feedback is an integral part of developing management information systems. A field experiment was conducted to study the process of el iciting user feedback during the design of a new financial system. Two variables were investigated--presentation of a system document in "finished" versus "draft" appearance, and use of a structured (questionnaire) versus unstructured mode for obtaining response. Results show that (1) use of a questionnaire elicited feedback faster and with significantly higher quantity and total perceived quality, (2) there was no significant difference in quantity, perceived quality, or time to respond between draft and finished presentations, and (3) there was no statistical interaction between the study variables. INTRODUCTION systems I iterature, reference to various aspects of the subject are found interUser involvement is critical to the spersed throughout. A survey of this existdevelopment of successful management ing literature reveals that eliciting user information systems (DeBrabander & feedback (1) has several purposes of key Edstrom, 1977). Further, the relationship importance in determining system content between users and developers is most beneand form, and (2) is prevalent in all phases ficial when it is one of cooperation and and most methodologies of modern system mutual understanding (Duncan, 1974). development. Acknowledging these findings, modern practices and methodologies for systems analysis, design, and installation aim for Purpose of Eliciting User Feedback consistent user/analyst interaction (via interviews, meetings, work groups, review With regard to the purposes of user sessions, inspections, etc.). Consequently, feedback, several specific objectives for a series of user/analyst feedback loops is obtaining user feedback can be identified created, generating the information needed in the literature. to build systems (King & Cleland, 1971). While such practices are fast gaining in 1. Detection of Errors. IBM popularity, little empirical research has researchers have conducted extenbeen done to study the mechanics of these sive tests on program walkfeedback processes in an information throughs and inspections and have system environment (Ramsey & Atwood, concluded that errors can be effi1979). ciently and effectively detected by obtaining feedback from other anaAlthough the specific topic of el iciting lysts and programmers (Duncan, user feedback is not comprehensively 1974). They also suggest that if addressed by any one body of information simi lar feedback loops are set up