Decision Analysis Presentations
INFORMS San Antonio, Fall 2000

To preserve the integrity of the authors' work we have limited your ability to manipulate the presentations on this web site.  You can download the pdf files but you cannot edit or copy any of the presentations; when requested by the author printing has also been disabled.  We have chosen to do this to more closely approximate your access to the material as an audience member at a presentation.

San Antonio Sessions, Cluster Chairs: Robin Dillon and John Butler

Panel: Government Applications of Decision Analysis (SA01), Thomas A. Edmunds
Representation & Solution of Asymmetric Decision Problems (SA02), Prakash P. Shenoy
Practitioners' Award Competition (SB01),  Detlof von Winterfeldt
New Directions in Behavioral Research(SB02), Jean L. Kahwajy
Medical Decision Making (SC01), Scott B. Cantor
Decision Analysis Arcade (SC02), Dana Clyman
Marketing & Decision Making (SD01), Robin Dillon and John Butler
Computation & Graphic Layout (SD02),  Ross D. Shacter
The Balanced Scorecard (MA01), James L. Ritchie-Dunham
Integrating Real Options & Decision Analysis (MC01), Ronald A. Howard
Decision Analysis Society Awards Presentation(MD01), L. Robin Keller
Decision Analysis at Schlumberger (TA01), Gary A. Lundeen
Environmental Applications of Decision Analysis (TC01)  Kara A. Morgan
Decision Analysis & Portfolio Applications (TD01), Peter C. Anselmo
Risk Management (TE01), Vicki Bier

If you would like to obtain a free Adobe Acrobat Reader so that you may access the pdf versions of the presentations, please click here

Return to main page for Decision Analysis Presentations from National Meetings

 

 


 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Panel: Government Applications of Decision Analysis

Session: SA01
Date/Time: Sunday 08:30-10:00
Chair: Thomas A. Edmunds
Chair Address: Lawrence Livermore National Laboratory, Decision Sci. Group, 7000 East Ave. Livermore, Livermore, CA
Chair E-mail: edmunds2@llnl.gov
SA01.1 Panel: Government Applications of Decision Analysis
Because decisions in the government sector often require consideration of a wide range of non-economic objectives and must be acceptable to a diverse set of stakeholders, they can be more complex than decisions in the private sector. We will describe some of our experiences in introducing decision analysis techniques into government decision processes...


Return to Session Index

Representation & Solution of Asymmetric Decision Problems

Session: SA02
Date/Time: Sunday 08:30-10:00
Chair: Prakash P. Shenoy
Chair Address: University of Kansas, School of Bus., Summerfield Hall, Lawrence, KS 66045-2003
Chair E-mail: pshenoy@ukans.edu
SA02.1 Representing & Solving Asymmetric Decision Problems (137K)
We present a formal framework called 'asymmetric influence diagrams' based on influence diagrams that allows an efficient representation of asymmetric decision problems. We also give an algorithm for solving asymmetric influence diagrams by decomposing the problem into symmetric subproblems.
SA02.2 Asymmetry in Decision Making in Practice
We address the prevalence, significance and variety of asymmetry in decision-making in practice in a number of applications. In particular, we provide examples from decision problems in project management and credit and risk management and identify asymmetries peculiar to these applications.
SA02.3 Representing Asymmetric Bayesian Decision Problems using Belief Functions (144K)
By viewing asymmetry as an uneven numerical specification of probabilities and utilities, a belief functions provide a most natural and compact representation of asymmetric decision problems. It avoids dummy events and acts, and degenerate probabilities and utilities. It also takes full advantage of numerical coalescence.
SA02.4 A Note on Asymmetry in Decision Problems (807K)
Several methods have been proposed for representing and solving asymmetric decision problems. However, it is not clear whether these methods are capable of representing every asymmetric problem. We study several asymmetric decision problems and examine their representation and solution using existing methods.


Return to Session Index
 

Practitioners' Award Competition

Session: SB01
Date/Time: Sunday 10:15-11:45
Chair: Detlof Von Winterfeldt
Chair Address: University of Southern California, Sch. of Policy/Planning/Dev., Los Angeles, CA 90089
Chair E-mail: detlof@aol.com
SB01.1 Practioners' Award Competition

 

 
 
 

Each year, the Decision Analysis Society of INFORMS presents an award for the best application of decision analysis. The finalists for the award will each present a summary of their work with the winner announced in the afternoon awards session.
 

Return to Session Index

New Directions in Behavioral Research

Session: SB02
Date/Time: Sunday 10:15-11:45
Chair: Jean L. Kahwajy
Chair Address: International Institute for Management Development, 23 Chemin de Bellerive, PO Box 915, Lausanne, CH-1001 , Switzerland
Chair E-mail: kahwajy@imd.ch
SB02.1 Subjective Probability Judgments: Partition Dependence (55K)
We present evidence that subjective probabilities depend critically on how the event space is partitioned. Our results suggest that judges compromise between 'ignorance prior' probability (equal mass across partitions) and probability based on support for the target event (as in support theory). We discuss implications for probability elicitation in practice.
SB02.2 Measuring Financial Investor's Risk Aversion (154K)
There is also an Excel spreadsheet with macros to aid in assessment available upon request from the authors.
We seek to devise a method for measuring an investor's risk aversion that fulfills 2 requirements: it asks preference questions meaningful to the individual and the resulting measure will translate into financial portfolio selection in a theoretically sound fashion.
SB02.3 Consulting Experiences in Decision Analysis
The hard part of decision consulting is integrating ideas, analysis and implementation with diverse personalities and competing interests. Some tricks of the trade have been discovered and will be discussed.
SB02.4 How to Hear & How to be Heard
The hard part of decision making is interpersonal communications. This research concerns the role that individuals play in reversing negative situations. It offers a radically different approach to negotiations and a productive strategy for being heard and for overcoming prevailing erroneous or negative expectations.
 

Return to Session Index

Medical Decision Making

Session: SC01
Date/Time: Sunday 13:15-14:45
Chair: Scott B. Cantor
Chair Address: University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Box 40, Houston, TX 77030-4095
Chair E-mail: sbcantor@mdanderson.org
SC01.1 Impact of Patient & Provider Preferences on the Cost-Effectiveness of Therapy for Recurrent Rectal Carcinomas
We performed a cost-effectiveness analysis of therapeutic options for patients with locally recurrent rectal carcinoma. We created a decision-analytic model that incorporated outcomes of survival, quality of life and costs. Utilities were elicited from convenience samples of 24 health care providers and 24 patients using the standard gamble technique.
SC01.2 Time-Tradeoff Utility Assessment with Equal Horizons: Would You Prefer 'Good then Bad' or 'Moderate'? (59K)
I propose a variation of the time trade-off method for utility assessment. Patients are asked to choose between a moderately unhappy life or a life of the same length that is first happy, then unhappy. Adjustments for temporal discounting can be made. Data from 50 patients will be presented.
SC01.3 Decision Making for the Treatment of Early Stage Prostate Cancer
Predictive modeling is commonly used in business decisions, occasionally in clinical research and rarely in the clinic. The use of modeling is highly sensitive to probabilities, utilities and assumptions, often supporting contradictory positions. Nonetheless, decision analysis for treatment of early stage prostate cancer could be promising for selective patients.
SC01.4 Prostate Cancer Screening Recommendations based on Couple's Utilities
We incorporated patient ages and husbands', wive's and couples' utilities into a decision-analytic model of prostate cancer screening. Subjects were 168 couples from 3 family pracice centers. Screening was recommended for 48 husbands, 89 wives and 58 couples. The recommendations showed greatest concordance for men's and couple's utilities.
 

Return to Session Index

Decision Analysis Arcade

Session: SC02
Date/Time: Sunday 13:15-14:45
Chair: Dana R. Clyman
Chair Address: University of Virginia, Darden Grad Sch. of Bus. Adm., PO Box 6550, Charlottesville, VA 22906-6550
Chair E-mail: clymand@darden.virginia.edu
SC02.1 Recent Decision Analysis Applications in Operations Research Literature (14K)
We survey applications of decision analysis that appeared in major English language OR journals and other closely related journals from 1990 through 1999. This is an update to the earlier survey of decision analysis applications from 1970 through 1989 by Corner & Kirkwood.
SC02.2 What's Cooking at the Lab?: Resource Allocation Decisions in Basic Science
The federal government will spend over $20 billion on basic science research in 2001. These resources are typically allocated via informal processes that allow emotion, politics and personal biases to play large roles. A decision analytic approach adds structure, insight and clarity to the process.
SC02.3 A Measure of Relevance (73K)
We introduce a single real number measure of relevance (probabilistic dependence) between discrete uncertain variables. The measure is easy to compute, is consistent with our intuitive assessment of relevance, can enhance the representation of relevance in influence diagrams and can improve communication between participants in the decision process.
SC02.4 (Most) Disagreements among Experts are Illusory
What seems to be a disagreement among experts about the state of the world is often just a result of vague language, reference to different questions or partial information. A more interesting source of apparent disagreement is a common methodological failure with regard to the role of experts in the evaluation of uncertain situations.
 

Return to Session Index

Marketing & Decision Making

Session: SD01
Date/Time: Sunday 15:00-16:30
Type: Sponsored
Sponsor: Decision Analysis Society
Chair: Jianmin Jia
Chair Address: Chinese University of Hong Kong, Dept. of Mktg., Shatin, NT, , Hong Kong
Chair E-mail: jjia@cuhk.edu.hk
SD01.1 Retail Loyalty Programs
The latest retail fashion: e-commerce, has eclipsed the last one: loyalty programs. Are they working? How are they supposed to work? The first half of the presentation will be ruminations on the issue, the second half will be results from a simple model of the situation.
SD01.2 Learning-Based Theories of Reputation-Building in Repeated Games: Scaring Competitors & Building Trust
We extend adaptive theories of learning in repeated one-shot games to games in which one 'teacher' plays with a series of learner students, so the teacher has an economic incentive to sacrifice short-run payoffs to teach what strategies do not pay. We use 2 examples: building a reputation for toughness by fighting entry attempts and building a reputation for trust by repaying loans.
SD01.3 Consumer Regret Following Switch vs. Repeat Deicsions in Outcome Sequences
Decision-making literature has consistently reported that decisions to maintain the status quo tend to be regretted less than decisions to switch. We examine the consequences of repeat purchases (maintaining the status quo) vs. switching in the context of information about prior consumption experiences, arguing that there are also situations where regret may be greater in the case of repeat purchases.
SD01.4 Consumer Preference Uncertainty: Measures of Attribute Conflict & Extremity
We investigate preference uncertainty as a function of stimulus characteristics such as attribute conflict (discrepancy among the attributes of an alternative) and attribute extremity (very high or low attribute values). Based on a random additive multi-attribute utility model, we derive formal measures of attribute conflict and attribute extremity and test our measures empirically using consumer purchase contexts...
 

Return to Session Index

Computation & Graphic Layout

Session: SD02
Date/Time: Sunday 15:00-16:30
Chair: Ross D. Shachter
Chair Address: Stanford University, Dept. of MS & Eng., Serra House, Stern Hall, Stanford, CA 94305-4026
Chair E-mail: shachter@stanford.edu
SD02.1 A Qualitative Linear Utility Theory for Spohn's Theory of Epistemic Beliefs (211K)
We formulate a qualitative 'linear' utility theory for lotteries in which uncertainty is expressed qualitatively using a Spohnian disbelief function. We argue that a rational decision maker facing an uncertain decision problem in which the uncertainty is expressed qualitatively should behave so as to maximize 'qualitative expected utility...'
SD02.2 Micro-Dynamic Simulataion for Information Gathering
One of the critical parts of the decision analysis is information gathering. In the world of new products and services development, there is information even beyond the expert judgement/opinion, or the situations where the expert opinions cannot be easily substantiated. We discuss micro-dynamic simulation for information gathering for decision making and the evaluation/quantification of expert opinions...
SD02.3 An Influence Diagram Model for Breast Cancer Screening (3284K)
Mammography is the best diagnostic technology currently available to decrease the mortality and morbidity of breast cancer, but mammographic interpretations and subsequent decisions vary widely among radiologists. We are developing influence diagram tools based on the standardized BI-RADS lexicon for mammography reports to improve medical decision-making.
 

Return to Session Index

The Balanced Scorecard

Session: MA01
Date/Time: Monday 08:15-09:45
Chair: James L. Ritchie-Dunham
Chair Address: University of Texas, Dept. of MSIS, 3615 Aspen Creek Parkway, Austin, TX 78749
Chair E-mail: jimrd@sdsg.com
MA01.1 The Theory behind the Balanced Scorecard (155K)
The BSC presents an attractive concept for measuring and managing organizational performance. We present current theoretically ungrounded BSC 'best practice,' as well as relevant theories that could strengthen it, such as multiple criteria decision making, means-ends analysis, statistical cause-effect methods, system dynamics and ERP systems.
MA01.2 Designing a Balanced Scorecard (129K)
A BSC requires both a scorecard and balancing. Creating a scorecard requires selecting elements to score and measures to describe performance on elements. Balancing requires evaluating different performance levels on an element and equating levels across different elements. Value-focused thinking and multi-attribute utility provide conceptual foundations for these requirements.
MA01.3 Influence Diagram Models for the Balanced Scorecard (31K)
The BSC approach to measuring and managing corporate strategy lacks a specific methodology for implementation. Influence diagrams bring a solid theoretic foundation to the representation of the structure and numbers that capture the causal relationships and dependencies needed to implement the BSC.
MA01.4 The Dynamic Scorecard: Objectives, Policies & Resources
The BSC 'operationalizes strategy,' suggesting a balanced policy set for achieving organizational goals over time. Developing policies about the accumulation and utilization of strategic resources over time requires dynamic thinking. This presents a forum for integrating periodic, value-focused decision making with policies about making decisions - DA meets SD.
 

Return to Session Index

Integrating Real Options & Decision Analysis

Session: MC01
Date/Time: Monday 13:15-14:45
Chair: Ronald A. Howard
Chair Address: Stanford University, Dept. of MS & Eng., Terman Eng. Ctr., Sch. of Eng., Stanford, CA 94305-4023
Chair E-mail: rhoward@stanford.edu
MC01.1 Real Options & the Black Scholes Formula: What's Wrong with this Picture? (166K)
We aim to reinforce the significance and value of real options by weeding out the most pernicious misconception associated with them, which is the presumed connection between real options and the Black Scholes formula, and point towards preferred alternatives for evaluating real options on solid quantitative and intuitive grounds.
MC01.2 Development Options
Most real options situations consist of a period of development of an opportunity leading to a launch event followed by a period of exploitation. Examples range from classic R&D, to creating new businesses, to making movies, to exploring for natural resources. I will present a general paradigm for characterizing development situations and several modeling and solution methods for evaluating them.
MC01.3 Real Options in Negotiation & Agreement Design
I present a framework for uncovering, structuring and evaluating real options in the course of a negotiation and in the life of an eventual agreement. Using a case on technology licensing agreements, I discuss the buyer's reservation line for negotiating contractual real options and the option value of negotiation.
 

Return to Session Index

Decision Analysis Society Awards Presentation

Session: MD01
Date/Time: Monday 15:00-16:30
Chair: L. Robin Keller
Chair Address: University of California, Grad. Sch. of Mgmt., 350 GSM, Irvine, CA 92697-3125
Chair E-mail: lrkeller@uci.edu
MD01.1 Decision Analysis Society Awards Presentation


Each year, the Decision Analysis Society of INFORMS presents awards for the best publications, the best student paper and the best application of decision analysis. The Society also periodically awards the Ramsey Medal for lifetime contribution; Dr. Detlof von Winterfeldt will be this year's Ramsey award recipient. Current awardees will be honored and will make presentations related to their work.
 
 
 

Return to Session Index

Decision Analysis at Schlumberger

Session: TA01
Date/Time: Tuesday 08:15-09:45
Chair: Gary A. Lundeen
Chair Address: Schlumberger, 8311 North FM 620, Austin, TX 78726
Chair E-mail: lundeen@austin.apc.slb.com
TA01.1 Supply Chain Management at a Tester Manufacturing Facility
We developed a LP-based tool for managing the 4-stage supply chain at a tester manufacturing facility. In the presence of long lead times, capacity restrictions and yield losses, this tool was useful for fast computation of ship dates. In addition, it enabled determining effective reactive (to meet quoted shipping dates) measures to uncertainties in the supply chain.
TA01.2 A Study in Risk Attitudes in Schlumberger Geco-Prakla Marine
Geco-Prakla Marine, like all companies, is regularly confronted with the issue of allocating scarce capital and resources among a set of available investment opportunities - opportunities generally characterized by some degree of financial risk and uncertainty. This study was conducted to determine decision maker risk attitudes in order to assist Geco-Prakla Marine management to develop their corporate risk policy.
TA01.3 Life Service Valuation as a Real Options Derivative of Marginal Oil-Field Value
Secondary recovery in maturing oil-fields can be aided by lift services. Many current onshore US fields are either at this stage of their life or will be in coming years. These fields have been an area for electrical submersible pump (ESP) lift services. While an increasing oil-price may make these fields more attractive for ESP services, possible future price downturns carry considerable revenue risk for the service provider...
TA01.4 Simulation of Back Deck Operations on a Marine Seismic Vessel
Marine seismic vessels deploy exploration equipment from their back deck in configurations called streamers. Streamer deployment in a new area, called mobilization, is a complex process and typically takes 4-5 days. This time is considered part of the operational overhead, and is not billed to the clients. Since revenues are measured in tens of thousands of dollars per day, reducing deployment times directly affects profit...
 

Return to Session Index

Environmental Applications of Decision Analysis

Session: TC01
Date/Time: Tuesday 13:00-14:30
Chair: Kara M. Morgan
Chair Address: Research Triangle Institute, 1516 M St. NW, Ste. 740, Washington, DC 20036
Chair E-mail: kmorgan@rti.org
TC01.1 Value of Information on Lower Trophic Level Uncertainties for Lake Erie Fisheries Management
The recent zebra mussel invasion has magnified uncertainties concerning phosphorus and energy flows into the management of Lake Erie's fisheries. Probabilities and multicriteria weights were elicited from experts from the Great Lakes Fisheries Commission, who then evaluated the expected value of information from research programs designed to reduce those uncertainties.
TC01.2 Stakeholder Goals for Water Quality Cleanup: Using Decision Analysis Tools to Help Articulate Public Values (71K)
The Neuse River suffers from nitrogen pollution and a modeling effort has been initiated to guide water quality regulations. To ensure that the models produce output meaningful to the public and decision makers, we elicited goals for river restoration from stakeholders. These fell into 4 substantive categories: biophysical, cultural, economic and recreation/health and 3 procedural categories: citizen involvement, fairness and balance...
TC01.3 Selecting a Remedial Action for a DNAPL Site
We will describe a multiple-stakeholder, multiple-objective decision analysis approach used to select a preferred cleanup alternative for a large site contaminated with DNAPL. Parties to the process included the responsible parties, the regulatory agency and a developer with plans to build on the site.
TC01.4 Understanding How Data Quality Impacts Decision Quality: Focus on Environmental Measurement Data (42K)
Science-based decision-making is grounded in theory, follows accepted methodologies and uses good data. The link between data quality and the decision is not always clear, especially to those doing the data collection. We present a proposed framework for understanding and communicating how data quality can (and should) impact the decision-making process.
 

Return to Session Index

Decision Analysis & Portfolio Applications

Session: TD01
Date/Time: Tuesday 14:45-16:15
Chair: Peter C. Anselmo
Chair Address: New Mexico Institute of Mining & Technology, Dept. of Mgmt., Socorro, NM 87801
Chair E-mail: anselmo@nmt.edu
TD01.1 Utilizing Portfolio Management in the Corporate Planning Process: A Case Study (163K)
Modern portfolio theory has provided managers in the petroleum industry a deeper understanding of the risk and returns associated with alternative mixes of assets. More and more companies have adopted this technique to gain better insights about their portfolio of E&P investment activities. We discuss one such application at a major oil company and discuss the implications of utilizing this approach...
TD01.2 Real Options & Manufacturing Capacity Decisions (41K)
We focus on a manufacturing plant that is projected to have excess capacity. We could potentially modify that plant so that it could produce one of several possible vehicles with that excess capacity. We address this problem using the theory of real options.
TD01.3 Valuation of Enhanced Oil Recovery Research
A portfolio optimization approach is used to value enhanced oil recovery research from the perspective of the funding firm. The basic idea of valuing enhanced recovery research as the expected present value of the difference between using or not using enhanced-recovery techniques is expanded to the context of a portfolio of drilling projects.
 

Return to Session Index

Risk Management

Session: TE01
Date/Time: Tuesday 16:30-18:00
Chair: Vicki Bier
Chair Address: University of Wisconsin, Mech. Engineering Bldg., 1513 University Ave., Rm. 451, Madison, WI 53706
Chair E-mail: bier@ie.engr.wisc.edu
TE01.1 A Risk Analysis for the Management of Safety & Delays in Airline Maintenance Operations (80K)
We studied the maintenance of the flaps and slats of an airplane leading edge for a major airline. We constructed a probabilistic risk analysis model, extended to include the maintenance process and quality as a function of factors such as location, crews' training and availability of spare parts. We then made recommendations for a more effective management of maintenance and for safer flights.
TE01.2 Value Focused Thinking, PRA & Supercomputers for Space Mission Design (921K)
The power of supercomputers now makes it possible to bring together value focused thinking and probabilistic risk assessment in simulation modeling for space mission design. The mission environment (space trajectories, and the surfaces of planets, asteroids and comets) and the mission systems can be simulated with probabilistic reasoning and sufficient validity to support major space mission decisions.
TE01.3 Designing Effective Risk Management Decision Processes (12K)
Many social risk management processes depend on information and participation of multiple stakeholders. Private information and incentives can distort production of relevant risk information, creating ineffective risk management decisions. We apply lessons from mechanism design theory and practice to suggest principles for designing more effective risk management processes.
TE01.4 Empirical Validity of Exponential Discounted Utility for Decisions involving Sequences of Monetary Outcomes
Experimental results on individuals' preferences for temporal sequences of monetary outcomes are discussed and compared to results on preferences for outcome/timing pairs. Anomalies that have surfaced in experiments with isolated outcomes (gain/loss asymmetry, long/short asymmetry and the absolute magnitude effect) are investigated with sequences of monetary outcomes over time.
 

Return to Session Index