objects/multiattribute-decision-making.html

Multiple Attribute Method for

Decision Making

This approach requires high information availability

statements of goals and objectives

honest statement of values

well managed process by knowledgable mediators

How to handle different dimensions of value

• public decisions are often conflicts of values
  • if they are decided one-by-one there is a high cost
  • shift to debate this at the level of rules for making the decision
• there are traditional ways of making decisions, called "folkways"
  • don't work as well as they could
  • see appendix
• multiattributes - bring in many different goals and values
• utility measurement
  • experimental, observational, judgement or combination
  • weighting factors - requires judgement at this step, not just at theend
• describe a 10 step method
  • see list
• Case study showed that using more attributes and this method allowed substantial agreement (even with weighting) whereas intuitive judgement focused on the few divisive differences in values
  • the differences don't necessarily go away, but they are put in a large context and how the multiple benefits

Ten step method

Ten Step - Simple Multi-Attribute ?? Technique SMART

1. Identify whose utility is being maximized
2. Identify issues, i.e. problems and alternatives
3. identify possible outcomes to be evaluated
   1. if action leades to uncertain outcome, may have to evaluate action
4. identify dimensions of value, often ignore the hierarchy and make a simple list of goals, keep list short
5. rank dimensions of values using discussion group input
6. rate importance of dimensions relative to each other
7. sum importances and normalize to 100 (8 dimensions ar plenty, 15 is too many, just a guiding rule)
8. measure the location (on X= dimension vs. Y = utility good for number of people)
   1. 3 classes of dimension (purely subjective, purely objective, mixture)
   2. subjective - get expert rating
   3. can identify differences of opinion between users
   4. straight ling between min & max plausible values
   5. develop utility curve for each dimension
   6. <!-- disucssion of linear vs other, i.e. threshold -->
9. sum utilities - weighted
10. decide - single shot or set of actions

Case Study on coastal development

competition for resources in the zone

California zoning commission set goals but no specific method to meet the goals

often, in testimony, values are stated as conclusions

group asked to process 15 hypothetical permit requests

crucial aspect of the permits

• size
• distance to mean high tide
• density
• building height
• unit rental rate
• land use in vicinity
• aesthetics

Evaluation and initial disagreement

individuals evalue scenarios and they provide a characterization of themselves on a range of conservation to development

initital, intuitive judgement resulted in some modest disagreements over ranking of the different permits between environmentalists vs. developers

using SMART - the disagreement was much less, in fact it was a substantial agreement

when using holistic/intuitive judgement to gropus focus on the points that engage their strong biases, but using SMART it cues them to look at many more dimensions

multidimensional analysis doesn't solve the disagreement on a few dimensions that are the most contentions, but adds other relevant dimensions that the group may agree on and may provide the basis for tradeoffs

2 types of disagreements

at step 8 - if they are not too large, can throw out the outliers and focus on the ones that are identified by best available experts

at steps 5 and 6 - it's a different kind of disagreement. This is a level of conflict of values. Decision should be made by decision maker or well chosen representative. If no decision is made, then the different values need to be passed all the way through the process.

SMART is a technology - a process to set up decision rules and then debate the problem at that level, rather than the specific decision. A more general approach allows bringing in more voices and perspectives.

References:

gardiner-edwards-1975.html

Gardiner, P. D., and Ward Edwards (1975). Public Values: Multiattribute-Utility Measurement for Social Decision Making. Human Judgement and Decision Processes. Martin F. Kaplan and Steven Schwartz, eds.. New York, NY, Academic Press, Inc: 1-38.

Appendix: Folkways for making decisions - currently in use

1. passionate advocacy - biased facts, focus on persuasion to a single point of view
2. Stages of evaluation - different methods for each stage, leads to change in character of the problem at each stage
3. Env Impact Report, is a necessary feature but ..., sets of numbers and statistics, may lead to stalling for more research, may provide a lot of data that isn't actually useful, doesn't state values
4. mushroom, keep them in the dark as much as possible
5. Agenda management, hearings and public meetings controlled to have important issues brought up last or on inconvenient dates, using parlimentary procedures as uneven-handed tool, exploiting loopholes in the procedures