Design and Implementation of Decision Support Systems: Some Implications of a Recent Study

By Charles B. Stabell
1975| Working Paper No. 252

This paper is an evaluation and critique of the Decision Support System (DSS) design and implementation methodology proposed by Gerrity (1970, 1971). A recent study (Stabell, 1974) of a system for portfolio management which was produced using Gerritys methodology forms the empirical base for the analysis. An initial review of Gerritys methodology identifies the key notion of change implicit in the development methodology: the purpose of a DSS development effort is to make possible and bring about a change towards more effective decision making behavior; the change is defined by the gap between the existing and a normatively prescribed decision making behavior; the change is assumed to come about by an evolution from support of the existing decision making process to support of the desired decision making process. Experience with the portfolio management system suggests that it is difficult to achieve the desired change in behavior as a result of a development effort based on Gerritys methodology. Three main, interdependent causes are identified in the critique: (1) low manager capacity for change which is not explicitly recognized and dealt with by the methodology; (2) the non-operational nature of the desired change in decision making behavior which is traced back to the methodologys decision modeling framework; (3) a designer centered development methodology which accentuates the problems arising from low manager capacity for change and which leads to a manager-user manipulative development effort. The analysis leads to proposed extensions of Gerritys methodology. The behavioral distinctions between two theories of rational behavior - the descriptive theories of bounded rationality and the normative theories of unbounded rational behavior - is proposed as a decision modeling framework that can better guide and direct both designer and manager-user. An extended role of the manager-user in the development effort is proposed as a means for dealing both with the managers limited capacity for change and with, the undesirable effects of a designer centered methodology. The thrust of the prescription is the notion that designers should design the DSS with the manager, not for the manager.