Strategic Decision Making Lectures
Lecture 1
Decision making: the process of selecting an alternative from a set of mutually exclusive alternatives.
Focus of the course is really on the process of decision making.
Basic terminology
- Strategic decision making
- Consequences of the alternatives you choose, depend on the choice made by others.
- More generically, strategic decisions relates to the overall long term strategy of an
organization, in relation to its external environment, which therefore always entail risk
and/or uncertainty.
- Rational decision making
- Rational: selecting what is best (by using your reason)
- Irrationality: not selecting the best outcome
- To be able to select the best thing, you need to have preferences.
- To be able to have preference, we need to attach value to things.
- Rational decision making: selecting a best alternative from a set of mutually exclusive
alternatives.
Decision theory
- Positive/descriptive decision theory: study of how real decision makers actually take
decisions. These are behavioural models.
- Normative decision theory: study of how decision makers should take decisions. This is based
on certain normative standards, e.g. equality, justice and fairness
- Prescriptive decision theory: study of how decision making could take better decisions. This is
based on frameworks of optimality.
Different types of statements:
- Primitive statements
- Derived statements
- Empirical statements
Two components of rationality
1. Behavioural component: the act of selecting a best alternative
2. Latent component: based on a preference. Preference itself is hidden, cannot be directly
observed.
3. Preference becomes apparent in our choices/actions (revealed preference = position on an
issue).
,Types of decision making problems
- Decision making under certainty: the outcome is known for every alternative. The
consequences of every alternative are known a priori.
- Decision making under risk: there are multiple outcomes for the alternatives. The
probabilities for arriving at each of these outcomes are known/can be calculated.
- Decision making under uncertainty: there are multiple outcomes for some of the
alternatives. The probabilities for arriving at each of these outcomes are not known/cannot
be calculated.
- Decision making under structural ignorance: outcomes for alternatives are unknown. We
don’t even know the consequences, let alone the probability distribution of these outcomes.
Independency of Irrelevant Alternatives
- When someone has to choose between A and B and chooses A, and thereafter has to choose
between A, B and C and chooses B, violates the condition of independency of irrelevant
alternatives.
With cyclical preferences, there is no transitivity of the decision maker. The decision maker cannot
value his/her decision. Preferences are needed to measure values. Transitivity of preferences is a
necessary requirements for the existence of a value (or utility) scale.
Decision making under certainty
- Axiom 1: completeness, i.e. preferences should be complete. If there are three options, there
should be preferences among all three.
- Axiom 2: transitivity
,Decision making under certainty
- Theorem 1: Preferences need to be transitive
- Theorem 2: if a preference satisfies completeness and transitivity, then a best alternative
exists.
- Theorem 3: if a preference R is transitive and complete, then there exists a value/utility
function.
Revisiting rationality
- Rationality is selecting the alternative that has the highest value/utility.
- Rationality has nothing to do with self-interest or egoistic behaviour. Also love and altruism
can be perfectly rational.
- Rationality has nothing to do with morality and ethics
- It merely presumes the existence of preferences that can be numerically represented (value
function or utility function)
Decision making under uncertainty
- Likelihood of situations is useful
Regret rule: minimize the maximal regret
- In the example you would choose engineering because it yields the minimum amount of
regret, namely three on money.
- Regret rule could violate the assumption of independence of irrelevant alternatives
Maxmin rule: selecting the best of the worst. Very conservative rule.
Conclusions
- Decision making is a process, leading to the selection of an alternative
- It requires preferences and the ability to choose/select
- Rational decision making: selecting the best alternative
- Requirements: completeness and transitivity
- Decision making under certainty: consequences/outcomes of each alternative are known.
The decision is then based on weighing relevant criteria. An elementary decision making rule
is the regret rule
- Decision making under uncertainty: consequences/outcomes of each alternative can vary.
Decisions are based on weighing consequences. Elementary decision rule is the Maxmin rule.
, Lecture 2
Decision making under certainty
Different decision strategies
- Lexicographic strategy: hierarchy of different criteria.
- Elimination: there is a minimal score for certain criteria.
- Sequential decision making: satisficing.
Decision analysis
- Heuristics (certain rules of thumb) do not necessarily lead to bad decisions.
- Most heuristics are however non-compensatory: there is little room for trade offs
- Thus with little room for trade-offs, there is no ability to compensate weak with strong
attributes
Decision analysis
- In complex decision-making situations, it is worthwhile to invest time and effort in
structuring the decision making process thus enabling compensatory techniques -> Multi
Criteria Decision Analysis.
- There are several subtypes of MCDA
SMART
AHP
MCDA
- In MCDA, each course of action becomes a vector of scores on criteria or attributes
- Alternative = option (i.e. possible outcomes of the decision)
- Attribute = criterion, i.e. a relevant feature of an alternative that measures the performance
in relation to an overall objective.
- Value: score that is attached to each alternative/attribute combination
- Objective: overall goal, movement of the decision making problem
- Weight: score that expresses the relative importance of an attribute in relation to the other
attributes.
Steps in MCDA
1. Defining the purpose
2. Defining the alternatives and setting up a value-tree (structure of decision process)
3. Evaluating alternatives on each attribute
4. Assigning relative weights to the attribute
5. Aggregating the weights of the attributes in order to obtain overall evaluation
6. Selecting the alternative with the best evaluation
7. Performing sensitivity analysis
Lecture 1
Decision making: the process of selecting an alternative from a set of mutually exclusive alternatives.
Focus of the course is really on the process of decision making.
Basic terminology
- Strategic decision making
- Consequences of the alternatives you choose, depend on the choice made by others.
- More generically, strategic decisions relates to the overall long term strategy of an
organization, in relation to its external environment, which therefore always entail risk
and/or uncertainty.
- Rational decision making
- Rational: selecting what is best (by using your reason)
- Irrationality: not selecting the best outcome
- To be able to select the best thing, you need to have preferences.
- To be able to have preference, we need to attach value to things.
- Rational decision making: selecting a best alternative from a set of mutually exclusive
alternatives.
Decision theory
- Positive/descriptive decision theory: study of how real decision makers actually take
decisions. These are behavioural models.
- Normative decision theory: study of how decision makers should take decisions. This is based
on certain normative standards, e.g. equality, justice and fairness
- Prescriptive decision theory: study of how decision making could take better decisions. This is
based on frameworks of optimality.
Different types of statements:
- Primitive statements
- Derived statements
- Empirical statements
Two components of rationality
1. Behavioural component: the act of selecting a best alternative
2. Latent component: based on a preference. Preference itself is hidden, cannot be directly
observed.
3. Preference becomes apparent in our choices/actions (revealed preference = position on an
issue).
,Types of decision making problems
- Decision making under certainty: the outcome is known for every alternative. The
consequences of every alternative are known a priori.
- Decision making under risk: there are multiple outcomes for the alternatives. The
probabilities for arriving at each of these outcomes are known/can be calculated.
- Decision making under uncertainty: there are multiple outcomes for some of the
alternatives. The probabilities for arriving at each of these outcomes are not known/cannot
be calculated.
- Decision making under structural ignorance: outcomes for alternatives are unknown. We
don’t even know the consequences, let alone the probability distribution of these outcomes.
Independency of Irrelevant Alternatives
- When someone has to choose between A and B and chooses A, and thereafter has to choose
between A, B and C and chooses B, violates the condition of independency of irrelevant
alternatives.
With cyclical preferences, there is no transitivity of the decision maker. The decision maker cannot
value his/her decision. Preferences are needed to measure values. Transitivity of preferences is a
necessary requirements for the existence of a value (or utility) scale.
Decision making under certainty
- Axiom 1: completeness, i.e. preferences should be complete. If there are three options, there
should be preferences among all three.
- Axiom 2: transitivity
,Decision making under certainty
- Theorem 1: Preferences need to be transitive
- Theorem 2: if a preference satisfies completeness and transitivity, then a best alternative
exists.
- Theorem 3: if a preference R is transitive and complete, then there exists a value/utility
function.
Revisiting rationality
- Rationality is selecting the alternative that has the highest value/utility.
- Rationality has nothing to do with self-interest or egoistic behaviour. Also love and altruism
can be perfectly rational.
- Rationality has nothing to do with morality and ethics
- It merely presumes the existence of preferences that can be numerically represented (value
function or utility function)
Decision making under uncertainty
- Likelihood of situations is useful
Regret rule: minimize the maximal regret
- In the example you would choose engineering because it yields the minimum amount of
regret, namely three on money.
- Regret rule could violate the assumption of independence of irrelevant alternatives
Maxmin rule: selecting the best of the worst. Very conservative rule.
Conclusions
- Decision making is a process, leading to the selection of an alternative
- It requires preferences and the ability to choose/select
- Rational decision making: selecting the best alternative
- Requirements: completeness and transitivity
- Decision making under certainty: consequences/outcomes of each alternative are known.
The decision is then based on weighing relevant criteria. An elementary decision making rule
is the regret rule
- Decision making under uncertainty: consequences/outcomes of each alternative can vary.
Decisions are based on weighing consequences. Elementary decision rule is the Maxmin rule.
, Lecture 2
Decision making under certainty
Different decision strategies
- Lexicographic strategy: hierarchy of different criteria.
- Elimination: there is a minimal score for certain criteria.
- Sequential decision making: satisficing.
Decision analysis
- Heuristics (certain rules of thumb) do not necessarily lead to bad decisions.
- Most heuristics are however non-compensatory: there is little room for trade offs
- Thus with little room for trade-offs, there is no ability to compensate weak with strong
attributes
Decision analysis
- In complex decision-making situations, it is worthwhile to invest time and effort in
structuring the decision making process thus enabling compensatory techniques -> Multi
Criteria Decision Analysis.
- There are several subtypes of MCDA
SMART
AHP
MCDA
- In MCDA, each course of action becomes a vector of scores on criteria or attributes
- Alternative = option (i.e. possible outcomes of the decision)
- Attribute = criterion, i.e. a relevant feature of an alternative that measures the performance
in relation to an overall objective.
- Value: score that is attached to each alternative/attribute combination
- Objective: overall goal, movement of the decision making problem
- Weight: score that expresses the relative importance of an attribute in relation to the other
attributes.
Steps in MCDA
1. Defining the purpose
2. Defining the alternatives and setting up a value-tree (structure of decision process)
3. Evaluating alternatives on each attribute
4. Assigning relative weights to the attribute
5. Aggregating the weights of the attributes in order to obtain overall evaluation
6. Selecting the alternative with the best evaluation
7. Performing sensitivity analysis
