Chair’s Column. Decision Analysis Problems for Classroom Use Revisited

In my column last December, I proposed we develop a set of contemporary decision analysis problems for classroom use and share them with each other via the DAS web site. The gist of that column (I will be happy to send you a copy if you wish) is that the academics among us need additional realistic problems that utilize structuring tools such as influence diagrams, encourage formulations in terms of variables rather than solely in terms of trees, use graphical sensitivity-analysis tools such as tornado diagrams, and generally utilize the capabilities of PC-based decision analysis software. We need additional sequential decision problems that do not appear to be contrived simply to require the calculation of reversed conditional probabilities and additional multiattribute problems from the private sector. While these problems can be short and simple or long and elaborate, they should look realistic, have plausible story lines, and pay at least some attention to data acquisition. Clearly, we need contributions from consultants and practitioners, who deal with real problems daily, as well as from academics.

Since that column did not lead to the submission of any problems from readers (assuming there actually are some readers), I am hereby making good on my threat to write a problem for communal use myself. The problem that appears below is simple enough for most of it to be solved by hand, so it can be utilized prior to the introduction of software if desired, with later parts assigned as relevant topics and tools are covered. Note that it illustrates some aspects of the data acquisition process rather than simply presenting the NPV’s needed for the decision tree as "given." It asks for an influence diagram as well as a decision tree and requires sensitivity analyses, at least one of which would be very tedious without using software. Embellishing the basic problem or the set of questions provided is straightforward.

John Butler of UT Austin has kindly volunteered to handle the web site for these problems, the location of which (courtesy of Jim Dyer) will be http://www.bus.utexas.edu/~dyerj/DA_probs/. If you can provide a suitable problem that is not subject to copyright or confidentiality difficulties, please send it in machine-readable form to John. Ideally, send him the problem as a text message or as a file attachment (e.g., in Word or WordPerfect format) at the following e-mail address: jbutler@utxvms.cc.utexas.edu. If this is not feasible, you can mail your submission to John at the following address: John Butler, c/o Professor James S. Dyer, CBA 5.202 (B6500), Graduate School of Business, University of Texas at Austin, Austin, TX 78712-1175.

Solutions should be available from the authors for all problems submitted, although they need not be in machine-readable form. Depending on the response to this initiative, we may consider developing a separate web page for solutions, with suitable security to prevent students from illicitly obtaining solutions to their assignments. Each problem posted on our web site will include the name, e-mail address, and telephone number of the author (a) to give the author whatever credit this confers and (b) to enable interested readers to contact the author to get solutions, to comment on the problem, to ask for additional background information, etc.

If a few of us would each provide one problem having at least one or two of the characteristics cited above, it would benefit the teaching of decision analysis, and thus our profession, significantly. Please contribute a suitable problem if you are able to do so.