OR/MS Today - October 1997 - Issues in Education

OR/MS Today, October 1997

End-User Modeling

By Tom Grossman

A growing number of business schools are making large changes to their MBA OR/MS courses. The most visible change is the use of spreadsheets, but there is also an emerging philosophical change. In this article I will attempt to articulate a philosophy, and discuss how we might use it in teaching.

Over the last 10 years I've met lots of people with MBAs. When I tell them I work in management science, a disturbingly common response has been, "Oh yeah, the simplex method. Waste of time." I have never heard the response, "Oh yeah, that incredibly useful stuff that got me a promotion." The common starting point for major course revisions seems to be a desire to replace the first response by the second response.

There is a growing interest in teaching an OR/MS course that enables students to contribute value to their firm in a way they couldn't before they took the course. An underlying philosophy might be articulated as a desire to educate MBA students to be "end-user modelers," who when faced with an ill-defined business "mess" are able to devise a useful model, manage inadequate data, perform appropriate analysis, extract insights and use them to communicate, persuade and drive change.

This end-user model of modeling is very different than the consultant model that underlies education in OR departments, and end users are very different from OR experts.

Many business processes have yielded to the consultant model of OR/MS. In general, these business processes have very high value (to justify paying and minding OR consultants), are stable over time (allowing the consultants a non-moving target), and have a rich data stream. Many, many more business processes do not have these properties. It is these processes, where change is continual and data is thin, "where managers are managers and OR analysts are scared," where end-user modelers are uniquely qualified.

Before proceeding, a small dose of professional humility is in order. Hordes of managers run businesses without the benefit of OR/MS, make suboptimal decisions daily, and on the whole seem to do pretty well. Although "optimality" is an invaluable mathematical concept, it is of little value as a business concept. The challenge of MBA OR/MS education is to equip experienced business people with the ability to use data systematically to do better tomorrow than they did yesterday, not to teach them technical details of tools and optimality.

What should we be aware of as we try to teach end-user modelers? We can recognize that the ability to roughly compute the effects of a proposed change ("what-if" modeling) is a powerful skill that is at least as important as the ability to compute a solitary optimum. We must also recognize that the ability to obtain quick, rough insight on actions that are likely to improve the business can be more beneficial than the ability to appreciate (for example) the richness of sensitivity range reports from LP models. And we should remember that in contrast to most of our textbooks, practicing managers often find that relevant, accurate data is a scarce commodity.

If we subscribe to this end-user modeler philosophy, we need to broaden the definition of OR/MS. Perhaps OR/MS in business schools should be the craft of imposing order and extracting meaning from the vast jumble of raw data that is available to most businesses. This is powerful and important, and can be achieved only by doing modeling. This entails less emphasis on tools, and increased emphasis on the hard work required before and after a tool is used.

Roughly speaking, we might crudely categorize end-user modeling into five steps: 1. "modeling" — creating a useful model; 2. "data" — acquisition and massage thereof; 3. "analysis" — manipulating a model (data-modeling, what-if analysis and appropriate use of the traditional strength of OR/MS: optimization); 4. "insight" — extracting the lessons for the business; 5. "persuasion" — using insights and model results to effect change in the organization.

There are many ways to teach these steps, and our collective experience of how to do this is increasing rapidly. Here are suggestions (from many teachers) that may help. First, make sure you're teaching to the entire class, not just the top 5 percent. Use small modeling cases (e.g., Icebergs to Kuwait), where data is inadequate and multiple good models and solutions are possible. Eliminate the word "optimality" from your MBA vocabulary. Teach how to perform what-if analyses, and how to compute tradeoff curves (these are not trivial skills). Don't be boring. Eschew algebra and use spreadsheets, the essential vehicle for data management, computation, "what-if" analysis and graphs. Incorporate a module on giving effective technical presentations. Give project assignments. Go to Forum on Education sessions at INFORMS conferences. Write cases, with MBA-scale challenges, and solutions that don't emphasize optimality, with messy data problems (and share them with others). And remember, success breeds success. The first graduate who comes back a hero because of your course can inspire an entire class.

My colleagues who teach finance routinely report students coming back to say, "Thank you, I've used what I learned in your course to build my career." It is disappointing that such reports are not routine in OR/MS. I believe that we can change this.

Tom Grossman of the University of Calgary is a former OR consultant with Decision Focus Inc. of Mountain View, Calif.

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com

Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com