This preview shows page 1. Sign up to view the full content.
Unformatted text preview: en et al. (1983) applied decision tree analysis to a problem faced by the
management of a coal-fired power plant in evaluating and selecting particular
emission control equipment. Winter (1985) used the techniques in management union
bargaining. A number of researchers (for example, Newendorp, 1996; Hosseini, 83 1986; Grayson, 1960) have applied decision tree analysis and EMV to drilling
Often these problems consider only two possible outcomes, namely success and
failure. However, in some problems the number of possible outcomes may be very
large or even infinite. Consider, for example, the possible levels of recoverable
reserves a company might achieve from drilling an exploration well. Such a variable
could be represented by a continuous probability distribution. This can then be
included in a decision tree by using a discrete probability distribution as an
approximation. A number of methods for making this type of approximation have
been proposed in the literature, the most commonly referred to is the ExtendedPearson Tukey approximation (EPT). This approximation technique, developed by
Keefer and Bodily (1983), based on earlier work by Pearson and Tukey (1965), is
acknowledged to generate good approximations to a wide range of continuous
probability distributions. For an illustration see Goodwin and Wright (1991 p110).
As Keefer and Bodily acknowledge however, the EPT approximation does have
limitations. For example, it is not applicable when the continuous probability distribution has more than one peak or the continuous probability distribution is
highly skewed. Despite this, the technique is widely recognised as providing a useful
mechanism for generating an approximation for continuous probability distributions
(Goodwin and Wright, 1991 p110).
The prescriptive decision analysis literature does not provide a normative technique
for eliciting the structure of a decision tree (some behavioural analysts have proposed
influence diagrams as a useful tool for eliciting the decision tree structure from the
decision-maker; see Goodwin and Wright (1991 p118) for a full explanation).
Structuring decision trees is therefore a major problem in the application of decision
analysis to real problems and, clearly if the structure is wrong, the subsequent
computations may well be invalid. Following such observations Von Winterfeldt
(1980) notes that it is good decision analysis practice to spend much effort on
structuring and to keep an open mind about possible revisions. However, problem
representation, according to Goodwin and Wright (1991), is an art rather than a
science. Fischoff (1980) argues similarly: 84 “Regarding the validation of particular assessment techniques we know …
next to nothing about eliciting the structure of problems from decisionmakers.” (Goodwin and Wright, 1991 p115)
Many decision-makers report that they feel the process of problem representation is
perhaps more important than the subsequent computations (Goodwin and Wright,
1991 p117). Humphreys (1980) has labelled the latter the “direct value” of decis...
View Full Document
- Summer '14
- The Land