Selection of decision support software for end users from the large number of types on the market is difficult. Vendors describe their packages in similar terms, but there are substantial differences in features, user friendliness, vendor support, and cost. Yet a choice must be made, hopefully one which will provide a stable base for an extended time period. The investment in applications of DSS software, especially in user expertise, makes conversion a costly and painful affair.

The article focuses on selection of mainframe-scale packages for modeling problems larger than microcomputer tools can handle. Typical applications include projections over time of sales, costs, earnings, and financial factors across dimensions such as products, markets, locations, and distribution channels. A complete checklist of selection criteria is the starting point for discussions on important points. It is not, however, a cookbook] Samples of major topics include:

• Modeling: multidimensional and consolidation features, nonprocedurality, simultaneous equations, built-in and user-defined functions, time dimensions.

• User Friendliness: syntax, documentation, prompts and provision of an “expert level” model, data entry.

• Analysis: “What if,” sensitivity analysis, goal seeking, risk analysis.

• Data Management: Supporting file system, security.

• Communications: Data interchange with other packages.

The authors note that current microcomputers lack the memory and computational power for large problems. There will always be a gap between mainframe and microcomputer technologies. An analytical approach which helps identify appropriate use of each technology, along the lines of this article, would be valuable.

This article is very well written. It can be understood without prior knowledge of the field. This type of article would be of maximum value if it appeared early in the life cycle of a class of packages. A similar article on inquiry and reporting packages would also be of use.