Decision Making
Decision-making is based on information Information is used to:
Identify the fact that there is a problem in the first place Define and structure the problem Explore and choose between different possible solutions Evaluate the effectiveness of the decision
Value of Information
The value of information used in decision making is:
(value of the outcome with the Information) (value of the outcome without the Information)
Types of Decision
H. A. Simon classified decisions into
Programmed decisions Non-Programmed decisions
Classified according to the extent to which decision making can be preplanned These are the extremes of a continuous range of decision types
Programmed Decisions
Also known as Structured Decisions Characteristics
Repetitive, routine, known rules or procedures, often automated, can be delegated to low levels in the organisation, often involve things rather than people Examples - Inventory control decisions, machine loading decisions, scheduling.
Non-Programmed Decisions
Also known as Unstructured Decisions Characteristics
Novel, non-routine, rules not known, high degree of uncertainty, cannot be delegated to low levels, more likely to involve people. Examples - Acquisitions, mergers, launching new products, personnel appointments. The most common type of decision May be partially automated
Semi-Structured Decisions
Empowerment
Authority to take decisions is being delegated down the line especially in modern service industries This process is called empowerment and should enable an organisation to take a variety of decisions more quickly, thus providing a more flexible service
Empowerment
Decisions should be made:
At the lowest possible level, which accords with their nature As close to the scene of the action as possible at the level that ensures none of the activities and objectives are forgotten
Empowerment
Enabled by systems such as
Customer Relationship Management (CRM)
Gives call centre staff specialist knowledge about any customer Assists non-experts in making complex decisions
Expert Systems
Uncertainty
Uncertainty arises from incomplete information due to:
Incomplete forecasting models Conflicting data from external sources Lack of time Internal data on particular problem not collated
The uncertainty of an outcome is expressed as a probability
Rational Decision Making
The rational model of decision making is a mechanistic approach to decision making It assumes perfect knowledge of all factors surrounding the decision
Rational vs. Real Decisions
Users tend to explain their actions in terms of rational behaviour, whereas their actual performance may be governed by intuition rather than by rational analysis. Studies of managers at work have shown that there is a discrepancy between how managers claim to take decisions and their actual observed decision-making behaviour. Argyris and Schon
Payoff Matrices
The standard way to analyse simple decision problems These are constructed as follows:
Identify all available options Identify events which cause an outcome (states of nature) Estimate the likelihood of each state of nature Estimate the value/payoff of each outcome Determine the expected value for each option Choose the option with the highest expected value
Example
A company must decide on one of three development projects, A, B or C They have identified three possible events relating to market conditions that will effect this decision
Probability 60% 30% 10%
Event Boom Steady State Recession
Example
The profit and loss figures (potential payoff) for the three products under the possible market conditions have been forecast as:
D e c is io n Event Boom 60% S te a d y S ta te 3 0 % R e c e ss io n 1 0 % P ro je c t A +8M +1M -1 0 M P ro je c t B -2 M + 6M + 12M P ro je c t C + 16M 0 -2 6 M
Which one of the above projects should the company run?
Decision Criteria
In order to evaluate the alternatives, managers use a number of different criteria: Equally Likely
The consequences of each decision are summed and the result divided by the number of events Useful if probabilities are not known Determine the highest possible profit from each strategy and choose that with the highest overall profit - Usually high risk, but high gain
Maximax
Example
Decision Event Boom 60% Steady State 30% Recession 10% Project A +8M +1M -10M Project B -2M +6M +12M Project C +16M 0 -26M
Preferred Project is?
Equally Likely Maximax
Decision Criteria
Minimax
Choose that action with the smallest maximum possible loss, or the largest minimum profit. Low risk, low gain. Choose the most likely event and then choose the best strategy for that event. Low risk, low gain. Does not make full use of available information.
Maximum Likelihood
Example
Decision Event Boom 60% Steady State 30% Recession 10% Project A +8M +1M -10M Project B -2M +6M +12M Project C +16M 0 -26M
Preferred Project is?
Minimax Max Likelihood
Example
Decision Event Boom 60% Steady State 30% Recession 10% Project A +8M +1M -10M Project B -2M +6M +12M Project C +16M 0 -26M
Decision Criteria
Expected Value
A weighted average of all outcomes The weights are probabilities
EV = { P( outcomei ) payoff i }
i =1 N
Gives the average value of the decision if it were made repeatedly Uses all the information concerning events and their likelihood
Example
Decision Event Boom 60% Steady State 30% Recession 10% Project A +8M +1M -10M Project B -2M +6M +12M Project C +16M 0 -26M
Calculate EV for each option/choice
Project A (8M*0.6)+(1M*0.3)+(-10M*0.1) = 4.1M Project B (-2*0.6)+(6*0.3)+(12*0.1) = 1.8 Project C (16*0.6)+(0*0.3)+(-26*0.1) = 7.0
Preferred Project is?
Example 2
A lte r n a t iv e AA r n a t iv e AB r n a t iv e C lte lte
y y y O u tc o m e :P r o b P r o fit P r o b P r o fit P r o b P r o fit
O p tim is t ic 0 .2 5 0 0 0 0 .3 4 0 0 0 0 .1 3 0 0 0 M o s t L ik e ly0 .6 7 5 0 0 0 .5 7 0 0 0 0 .7 6 5 0 0 P e s s im is tic0 .2 9 0 0 0 0 .3 9 5 0 0 0 .2 1 0 0 0 0
Decision Criteria
Expected Value
Uses all the information concerning events and their likelihood Does not take into account decision-makers attitude to risk Does not reflect the actual outcomes in the figures
Can the company afford to lose 26M?
Decision Trees
Not all decisions will be taken in isolation A decision will have an effect of future events and outcomes An outcome in turn may effect future decision making
Decision Trees
Decision trees provide a means of structuring the decision making process to allow for alternative futures
Decision Tree
Two types of Node Decision Node
Represent decision points Decision are made by the organisation Linked to possible outcomes These are uncontrollable
Outcome Node
Example
Boom 60% Steady 30% Project A Boom 60% Project B Steady 30% Recession 10% Boom 60% Steady 30% Recession 10% -26M Recession 10% 8M 1M -10M -2M +6M +12M +16M 0
Project C
Example
Boom 60%
4.1
8M 1M -10M -2M +6M +12M +16M 0 -26M
Steady 30% Recession 10% Boom 60%
Project A
Project B
1.8
Steady 30% Recession 10% Boom 60%
Project C
7
Steady 30% Recession 10%
Example
Boom 60%
4.1
8M 1M -10M -2M +6M +12M +16M 0 -26M
Steady 30% Recession 10% Boom 60%
Project A
4.1
Project B
1.8
Steady 30% Recession 10% Boom 60%
Project C
7
Steady 30% Recession 10%
