Abbas recognized for two important papers on decision-making

Two papers co-authored by Ali E. Abbas, the Art Davis Faculty Scholar and associate professor in the Department of Industrial and Enterprise Systems Engineering, were recently recognized by the Decision Analysis Society at the INFORMS annual conference. The papers challenge the current state of the art making for tradeoffs in multi-objective decision problems.

“The idea of better understanding tradeoffs in the decision-making process has broad applications in numerous fields including global climate change, health care, design and manufacturing, education and other important fields,” explained Abbas, who directs the Information Systems and Decision Analysis Lab at Illinois. “The papers propose fundamental new machinery to help decision-makers express more accurate tradeoffs.

“To provide some insights into the research that was conducted, it is important to observe that the field of making trade-offs has relied on constructing simplified forms of multi-attribute utility functions that were simple to assess but seldom represented the true preferences of the decision makers,” Abbas added. These simplified forms relied on stringent conditions known as “utility independence” that have been used in the last four decades.

According to Abbas, this prior work required the decision-maker to make certain assertions that preferences between uncertain deals over an attribute do not change when the levels of another attribute vary. For example, in a medical decision with two attributes, health state and wealth, independence conditions would require assertions that preferences for investments over money do not change with our health state.

“In climate change, they would require assertions that preferences for investments in climate change would not change with the actual climate temperature. It is very unlikely that such preferences would hold in real life: preferences for the type of investments we make do in fact change with our health state or with climate change.”

In the first paper, “The Multiattribute Utility Tree,” Abbas provided a general method for making trade-offs in multi-objective decision problems and for thinking about the functional forms of multi-attribute utility functions without making utility independence conditions. He proposes an expansion theorem for multi-attribute utility functions that is analogous to Bayes’ expansion theorem for probability functions.

Using this general expansion theorem, Abbas derived more general functional forms of utility functions than those implied by utility independence, thereby allowing for more general tradeoffs and preferences to be modeled. He also shows that simplified functional forms of utility functions can also result without having to make the stringent assumptions of utility independence.

“The work will set a fundamental milestone for making tradeoffs defining the calculus of multi-attribute utility,” Abbas remarked.

In the second paper, “One-switch Independence for Multi-attribute Utility Functions”, co-authored with David Bell from the Harvard Business School, Abbas further relaxes the “utility independence” conditions. He proposes a new condition that still results in simplified functional forms of utility functions yet does not make stringent assumptions like utility independence. The new condition is called “one-switch independence.”

In their paper, Abbas and Bell ask, “What happens if preferences for investments may change with health state but only once (or twice or any number of times)? What does this imply about our tradeoffs? The authors derived the general functional forms of utility functions that satisfy these conditions thereby further generalizing the functional forms assumed by utility independence and allowing for richer tradeoffs.  

The two papers were recognized—from among numerous submissions from all over the world—as first and second runner-up in the Best Publication category (first place went to an edited book, rather than a research paper). Abbas previously received dual honors in 2011 for the Best Publication Award from the Decision Analysis Society in 2011. He also received the Best Publication Award for work on making tradeoffs in target-based settings, and he was the runner-up for Best Paper honors for work on relaxing independence conditions in that same year.

• Abbas, A.E., "The Multiattribute Utility Tree," Decision Analysis, 8 (3), pages 165-169, 2011.
• Abbas, A.E. and D.E. Bell, "One-Switch Independence for Multiattribute Utility Functions," Operations Research, 59 (3), pages 764-771, 2011.