Summary Decision Analysis for Management Judgment
Overview of the book
Chapter 2: biases that arise when unaided decision makers face decision problems involving multiple
objectives.
Chapter 3: then shows how decision analysis can be used to help with these sorts of problems.
Chapter 4: presents some alternative methods for handling decisions
Chapter 6: where we apply probability to decision problems and show how the decision maker’s
attitude to risk can be incorporated into the analysis.
Chapters 7: illustrates decision trees and influence diagrams
Chapter 9: how a decision maker should revise judgments in the light of new information.
Chapter 10: reviews psychological evidence on how good people are at using judgment to estimate
probabilities.
Chapters 13: describe problems that can occur in group decision making and discuss the role of
decision analysis in this context
Chapter 15: framing problems and the cognitive inertia
--------------------------------------------------------------------------------------------------------------------------------------
Chapter 1 Introduction
1.1 Complex decisions
Interrelated features of decision problems in general are:
Multiple objectives: problem involves multiple objectives
Uncertainty: the problem involves uncertainty
Complex structure: much of your frustration in attempting to understand your decision problem
arises from its complex structure.
Sequential: some of the decisions are sequential in nature
Multiple stakeholders: decision involves multiple stakeholders
1.2 The role of decision analysis
Decision analysis therefore involves the decomposition of a decision problem into a set of smaller
(and, hopefully, easier to handle) problems. Because decision analysis requires the decision maker to
be clear and explicit about his or her judgments, it is possible to trace back through the analysis to
discover why a particular course of action was preferred. The basic assumption is that of rationality.
If the decision maker is prepared to accept a set of rules (or axioms) which most people would regard
as sensible then, to be rational, he or she should prefer the indicated course of action to its
alternatives.
1.3 Good and bad decision outcomes
When we consider a single decision, the outcome usually provides, at best, only limited information
about the quality of the decision. Outcomes across many decisions provide a better guide. If decision
analysis is being used to support a decision, how should we assess its effectiveness? Schillingetal
suggest three main criteria:
1. Quality of the process: measured by such factors as the extent to which people in the
organization participated in the decision-making process, and exchanged information, the extent
to which the process was transparent and comprehensible and how much it yielded insights into
the problem.
2. Output effectiveness: embraces both ‘hard’ factors like increased profit and ‘softer’ benefits like
the provision of a common language.
3. Outcome effectiveness: which is usually more difficult to measure, relates to the long-term
consequences of the analysis.
,Chapter 2 How people make decisions involving multiple objectives
2.1 Introduction
This chapter looks at how decision makers make intuitive decisions involving multiple objectives.
Many decisions involve multiple objectives (cheap vs resort). Such intuitive decision making is usually
‘unaided’: by which we mean that people face decisions like this without the support and structure
provided by the decision-analysis methods.
Following terms used by Simon:
Bounded rationality: the fact that the limitations of the human mind mean that people have to
use ‘approximate methods’ to deal with most decision problems and, as a result, they seek to
identify satisfactory, rather than optimal, courses of action.
Heuristics: approximate methods or rules of thumb.
Fast and frugal heuristics: quick ways of making decisions, especially when time is limited.
2.2 Heuristics used for decisions involving multiple objectives
When a decision maker has multiple objectives the heuristic used will either be:
Compensatory: an option’s poor performance on one attribute is compensated by good
performance on others. E.g. high price for a computer with high processor.
Non-compensatory: this compensation is not the case in this strategy.
Non-compensatory theories:
The recognition heuristic: used where people have to choose between two options. If one is
recognized and the other is not, the recognized option is chosen (e.g. you know the brand of one
product and not of the other product). This strategy is non-compensatory, once an alternative is
recognized, no further processing of additional cues takes place. This strategy works well:
o In environments where quality is associated with ease of recognition
o Choices have to be made on how to rank objects on some criterion
The minimalist strategy: decision maker first applies the recognition heuristic, but if neither
option is recognized the person will simply guess which the best option is. If both options are
recognized then the person will pick at random one of the attributes of the two options.
Take the last: make the decision based on the past. Rather than picking a random attribute,
people recall the attribute that enabled them to reach a decision last time when they had a
similar choice to make (e.g. choose the last airline company whit which you flew).
The lexicographic strategy: decision maker ranks the attributes in order of importance (e.g.
choosing a car based on air-conditioning). This strategy involves little information processing (i.e.,
it is cognitively simple) if there are few ties. Also a non-compensatory strategy.
The semi-lexicographic strategy: the performance of alternatives on an
attribute is similar, the decision maker considers them to be tied and
moves on to the next attribute. If the price difference between brands
is less than 50 cents choose the higher quality product, otherwise
choose the cheaper brand.
Elimination by aspects (EBA): the most important attribute is identified and a cutoff point. Any
alternative which has a performance falling outside this boundary is eliminated. For example:
o Price is the most important attribute: eliminate all cars costing more than $15000 and
any costing less than $6000.
o Number of seats is the next most important consideration: eliminate two-seater cars.
o Etc.
Sequential decision making (satisficing): alternatives become available sequentially (e.g. looking
for a house to buy, month new houses available). In satisficing, decision makers stop searching as
soon as they find an alternative that is satisfactory. Important characteristics:
o Aspiration level: which characterizes whether an alternative is acceptable or not.
, o Change in aspiration level: decision makers’ aspiration levels may change during the
search process as they develop a better idea of what they can reasonably achieve.
Reason-based choice: when faced with the need to choose, decision makers often seek and
construct reasons in order to resolve the conflict and justify their choice to themselves and to
others. Often violates rational decision making. Principle of rational decision making is that of
independence of irrelevant alternatives: e.g. prefer holiday to French over Mexico should still
hold if a third holiday to Canada comes in place.
Factors affecting which strategies people employ
Factors that affect strategies people employ include:
1. The time available to make the decision
2. The effort that a given strategy will involve
3. The decision maker’s knowledge about the environment
4. The importance of making an accurate decision
5. Whether or not the decision maker has to justify his or her choice to others
6. A desire to minimize conflict (for example, the conflict between the advantages and
disadvantages of moving to another job)
2.3 Other characteristics of decision making involving multiple objectives
Decoy effects: these are options which asymmetrically dominate a particular option but then
turn out to be unavailable. E.g. you want to buy a laptop which is discounted for the 20 earliest
people, but you still buy it if your are not the first.
Choosing by unique attributes: the degree to which attributes are shared across alternatives
influences which alternatives are preferred.
Emotion and choice: emotion can also influence how information is processed. Sad moods tend
to prompt detailed analytic thinking, whilst happy moods tend to result in less-detailed analysis.
Justifying already-made choices: decision makers may, initially, overvalue the choices that they
have resolved to make (e.g. decided to buy a product they may overpay at the time of purchase).
Partitioning the total cost of an item changes preferences: the total payment for a product or
service can be presented in either partitioned ($199 plus $30 shipping) or aggregated ($229)
form. Research shown that display the product cost and the shipping cost separately is the best.
Chapter 3 Decisions involving multiple objectives: SMART
3.1 Introduction
When decision problems involve a number of objectives unaided decision makers tend to avoid
making trade-offs between these objectives. This might lead to the rejection of relatively attractive
options because their good performance on some objectives is not allowed to compensate for their
poor performance elsewhere. Therefore, this chapter explore how decision analysis can be used to
support decision makers who have multiple objectives.
, The methodology outlined in this chapter is underpinned by a set of axioms. We will discuss these
toward the end of the chapter, but, for the moment, we can regard them as a set of generally
accepted propositions or ‘a formalization of commonsense’. If the decision maker applies the axioms,
he/she will make decisions in a rational way.
3.2 Basic terminology
Objective: an indication of the preferred direction of movement. Thus, when stating objectives,
we use terms like ‘minimize’ or ‘maximize.’ (e.g. minimize pollution).
Attribute: is used to measure performance in relation to an objective.
Proxy attribute: use an attribute which is not directly related to the objective.
Value: if the decision involves no element of risk and uncertainty.
Utility: where the decision involves risk and uncertainty.
3.3 Simple Multi-attribute rating technique (SMART)
Simple multi-attribute rating technique (SMART). SMART has been widely applied because of its
relative simplicity and transparency. Example of a problem: a business has to move to a different
office. The main stages in the analysis are:
1. Identify the decision maker (or decision makers): e.g. the business owner.
2. Identify the alternative courses of action: e.g. the different offices that can be chosen.
3. Identify the attributes which are relevant to the decision problem: attributes of the offices.
4. For each attribute, assign values to measure the performance of the alternatives on that
attribute: e.g. how well do the offices compare when considering the quality of the working
conditions they offer?
5. Determine a weight for each attribute: this may reflect how important the attribute is to the
decision maker.
6. For each alternative, take a weighted average of the values assigned to that alternative: this will
give us a measure of how well an office performs over all the attributes.
7. Make a provisional decision
8. Perform sensitivity analysis: to see how robust the decision is to changes in the figures supplied
by the decision maker.
3.4 Constructing a value tree
We can start constructing the tree by addressing the attributes which represent the general concerns
of the decision maker (costs & benefits). We next need to break down the cost and benefits of the
offices into more specific attributes that will make it easier for us to compare the locations. Keeney
and Raiffa have suggested five criteria which can be used to judge the tree:
1. Completeness: if the tree is complete, all the attributes which are of concern to the decision
maker will have been included.
2. Operationality: this criterion is met when all the lowest-level attributes in the tree are specific
enough for the decision maker to evaluate and compare them for the different options.
3. Decomposability: this criterion requires that the attractiveness of an option on one attribute can
be assessed independently of its attractiveness on other attributes.
4. Absence of redundancy: if two attributes duplicate each other because they actually represent
the same thing then one of these attributes is clearly redundant.
