FROM THE BOOKSHELFANDREW RUPPEL,
Feature Editor,
McIntire School of Commerce, ARE YOU CRAFTY? ARE YOU ETHICAL?by Andrew Ruppel, University of Virginia Building models is at the core of decision science activity. Becoming more adept in the craft of model-building is a matter of personal pride and professional responsibility, as should be the desire to maintain an ethical posture. Here are two books by veteran modelers offering reflective insights into model-building practice and professionalism.
John Wiley & Sons, 1994. 304 pages. In the world of operational research (as the Brits call it), Patrick Rivett's name is well known. The first professor of operational research in the U.K., he has taught and consulted on both sides of the Atlantic. In this, his sixth book, Rivett offers his view of what makes for successful modeling. And the view that he takes is not that of a computational specialist or 'number cruncher,' but rather that of a 'number questioner.' A crafty modeler, says Rivett, ``should always start modelling by listening to people.'' Thus one needs to be friendly, extroverted, and maintain a humorous outlook. And this the author does as he interweaves fifteen chapters with nine case studies from the ``paddy fields of practice.'' This is not a traditional textbook, but it is eminently suitable collateral reading in graduate courses for specialists and non-specialists, students and managers alike. Rivett sees the emergence of soft systems methodology (i.e., more structured dialogue between modeler and client) and the rise of information technology as presenting new approaches to problem description and solution handling. However, he is dismayed that these two developments have not interlaced themselves more. Information technology has had the effect of enlarging the managerial span of control and collapsing the organizational hierarchy, though, as Rivett observes, ``organizations can be cybernetically unsound in the way they divide their operation into functional parts.'' The nine case studies are both descriptive and didactic. Within each study or 'Life' as the author oddly labels them, a beginning anecdote with accompanying moral leads to a more elaborate problem articulation, followed by a laid-out solution effort. The case studies can be read separately from the ten chapters that are woven among them. Indeed, they make fascinating reading in their own right. Each is conveniently summarized in Chapter 11. The remaining four chapters deal with practical, educational, and ethical dimensions of modeling activity. For more, much more, on the latter subject, see the following.
by William A. Wallace, editorElsevier-Pergamon, 1994. 266 pages. In 1989, a workshop was held at the Rensaleer Polytechnic Institute on ``Ethics in Modeling.'' Sponsored by the Sloan Foundation, the workshop addressed the concern that modeling activity was proliferating (which was good), but that its results were not always being responsibly handled (which clearly was bad). Workshop convener Professor William A. Wallace of RPI has now brought together the nine original presentations at the workshop with four additional papers by attendees, plus introductory comments of his own, to provide a compilation appropriate for use in graduate seminars. Practicing decision science professionals will also appreciate the views expressed and examples given, e.g., the court-directed breakup of AT&T, and the undercount adjustment controversy in the 1990 U.S. Census. Several ethics issues are recognized by the eighteen contributors. The first of these is lack of adherence to the replication principle of science: that modeling work should be conducted and documented in such a way that others can readily carry out confirmatory checks on the soundness of the methods used and the conclusions drawn. Modelers operating in a proprietary setting clearly have challenges in meeting this principle. A second and more prominent ethical modeling issue concerns how values are handled or mishandled within the model. These values include those of the client, the values of the modeler, the values of those affected (knowingly or unknowingly) by the model's recommendations, and the values of society. At another level is the concern for giving proper credit for intellectual contributions to the modeling effortþa concern of particular relevance in academia. (Interestingly, the name index for Ethics in Modeling is longer than the subject index.) Two of the contributors observed, perhaps somewhat cynically, that no operations researcher had ever had his professional society membership revoked for unethical modeling. Indeed, the most difficult question to answer is how any ethical modeling conduct code would be enforced. In attempting to offer some answers to the ethical-code question, Saul Gass in his paper presents the codes from several modeling 'arenas': the social sciences, the policy sciences, the health sciences, consulting engineering, the computing sciences, and operations research. Gass also provides an overview of the personal code employed by the late W. Edwards Deming in his statistical consulting. It has 37 points organized into 7 sections. Deming's personal code reveals that he was extremely careful about drawing a sharp boundary between where his statistical interpretation ended and management's use of it began. He would not specify the actions that management should take. This posture is contrary to his well-known aggressive advocacy of TQM principles. And one wonders how Deming would react to the remark by Warren Walker in his paper on responsible policy modeling that ``the question of `ethics in modeling' is really a question of quality control.'' The closing chapter makes a plea for more sensitivity and responsibility toward values in modeling. More societal involvement is also called for. The academic discipline of Applied Ethics should be brought into play, says the chapter's authors. All this sounds a bit too glib. The serious questions raised by the workshop's participants, hard as they are to address, are not answered by merely 'taking another college course.' Editor Wallace more realistically summarizes the situation better in his introductory chapter, wherein he says: ``Dealing with the issues of representing reality is far easier than attempting to translate values into the formalism of a model.''
Dr. Andrew Ruppel |