OR/MS Today - February 2000

February 2000

Knowing When to Say When

By Douglas A. Samuelson

"So how's your teaching going?" the bank manager asked as the OR/MS analyst turned away from the teller's window, receipt in hand. They had known each other for some time, and the analyst had shared some of his stories about what he did. The analyst rolled his eyes. "Mostly OK, Ann," he said, "but the few who don't get it really don't get it."

"This sounds like a story." Ann's eyes twinkled.

"I was teaching decision analysis last semester," the analyst explained. "The book, like most books for this subject, is heavy on tools and techniques, but a bit light on interpretation, and more than a bit light on ethical implications. I kept trying to fill in more on those subjects with mixed success. Some of the students understood that my experience in real business taught me some important lessons about the limitations of these decision methods. Others apparently thought those stories were distractions, wasting time that could have been spent better on how to do the calculations. I got back their evaluation forms a few days ago, and that's what they said. Some of the criticism probably is justified and really means I didn't present it well, but some of it sounds to me like people refusing to believe that decisions are subjective, no matter how we try to make them less so."

"That's what my business professors taught," Ann volunteered. "One of them told us that a quick, reliable way to make a hard decision between two pretty close alternatives was to flip a coin, cover it with your hand without looking at it, and then quickly pay attention to how you hope it came out. Usually you do have a feeling, one way or the other. That's your choice!"

"So much for decision trees and net present value analysis," the analyst laughed. "But that's not a bad way to choose a restaurant. I've done that with my kids.

"The thing is," he continued, "I was trying to teach these students that even when the decision is complicated, and tools like net present value or location analysis are useful, you still have to watch your assumptions. It's easy to end up structuring a decision so you're captured by your methods, even though the results don't make much sense!

"One example in the textbook I used," he went on, "was about choosing among three job offers. The person cares a lot about good winter sports, so she uses average annual snowfall in the three cities as a variable in her decision analysis. So she ends up choosing St. Paul, Minn., over Flagstaff, Ariz., and San Francisco.

"I gave them an exam question where I asked them to extend the example in the book and critique the criteria. A few of them were really mad at me for taking off several points if they didn't mention that snowfall is a lousy measure of access to winter sports! San Francisco has virtually no snow, but they're three-and-a-half hours from some of the best winter sports in the country. I used to work there, and those charter buses were full every weekend. St. Paul gets lots more snow, but it's pretty flat around there. I don't know how far you have to go from there if you want downhill skiing, but I'd bet it's a lot farther than Lake Tahoe is from San Francisco!

"Anyway, I did hear one really choice excuse, the excuse of the computer age, I guess. This one guy insisted that he had discussed those issues. I didn't see the discussion, so I took off some points. I agreed to go over the exam, but I still didn't see it. So then he took another look, and came up with the explanation that he had written about it, but then he pasted a graph into the document, and the graph overlaid that part of the text. In other words, Œthe cyber-dog ate my homework!'"

Ann joined him in a hearty laugh.

"But then he was still arguing, as well, that it wasn't important anyway," the analyst continued. "He was sure that the tools and techniques were all that really mattered, and why couldn't I see that? Of course, by arguing that way he just dug himself into a hole, since that Œreasoning' confirmed my impression that he had never gotten the point."

"So you're agreeing with the management writers who say these quantitative methods are useless?" Ann asked slyly.

"No," the analyst quickly demurred. "But they're an aid to judgment, not a substitute for it. And whenever you make a complex decision, you need to develop not only a plan of action, but also a plan to detect whether you've gone wrong. You need some intermediate milestones. Then, if your results are much different from what you expected, you need to reconsider whether your decision and the assumptions going into it were right, not just what could have gone wrong with the implementation.

"And, of course," he grinned, "you can also use sophisticated methods to help you clarify what's really important. Don Gross, one of my professors, claimed that the way to pick a spouse was to list, carefully, all the qualities you were looking for in a mate and then marry the first one who made you forget the list!"

Douglas A. Samuelson is the president of Infologix, Inc.

Table of Contents

OR/MS Today Home Page

Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com