DECISION-MAKING TOOLS
I. Models for decision making
-- Choice of models
 Real life experiments
 Iconic models
 Simulation models
 Quantitative models
> Algebraic - B/E analysis, cost/benefit analysis
> Statistical analysis – forecasting, quality control,
decision theory
> Linear programming - a variety of POM applications
> Queuing theory
> Inventory models
> Network models - PERT and CPM
Which model to use?
-- Advantages of using models
 less expensive and disruptive than real world experimentation
 permits “what if” type of questions and scenarios
 built for management problems and encourage management input
 force a consistent and systematic approach to problem solving
 require managers to be specific about constraints and goals
 help to reduce the time needed in decision making
-- Disadvantages of using models
 may be expensive and time consuming to develop and test
 often misused and feared because of their mathematical
complexity
 tend to downplay the role and value of non-quantifiable
information and qualitative reasoning
 assumptions sometime not realistic
II. The Decision Process
 Define the problem and the factors that influence it
 Establish decision criteria and goals
 Formulate a model or relationship between goals and decision
variables
 Identify and evaluate alternatives
 Select the best alternative
 Implement the decision
 Evaluate the results
III. Decision Theory
-- An example problem: Your T-shirt business makes a $10 profit
for each shirt ordered and sold, but loses $5 for each unsold
shirt.
-- Fundamental terms in decision theory
 alternative - course of action that must be chosen by the
decision maker
 state of nature - an occurrence over which the decision maker
has little or no control
-- Types of decisions
1. Decision making under certainty – the decision maker knows
the outcome for an alternative/decision with certainty
(probability = 1)
2. Decision making under uncertainty – the decision maker knows
possible outcomes but not the probability associated with
each outcome
Step 1. Create a decision table (payoff table)
 List alternatives along one axis and states of nature
along the other axis
 Write or calculate outcomes (payoffs) in the body of the
table
Demand 
Decision 
Order 10
Order 20
Order 30
States of Nature (Demand)
10
20
30
shirts
shirts
shirts
Step 2. Make the decision based on a criterion
 maximax – choose the alternative that has the best outcome
in the best case scenario, a very optimistic criterion
1) Find the maximum payoff for each alternative
2) Choose the alternative with the largest maximum

maximin - choose the alternative that has the best outcome
in the worst case scenario, a very pessimistic criterion
1) Find the minimum payoff for each alternative
2) Choose the alternative with the largest minimum

equally likely - choose the alternative with the highest
average outcome
1) Find the average payoff for each alternative
2) Choose the alternative with the highest average

minimax regret – choose the alternative with the least
opportunity cost (the largest regret), another pessimistic
criterion
Demand 
Decision 
Order 10
Order 20
Order 30
States of Nature (Demand)
10
20
30
shirts
shirts
shirts
$100
$50
$0
$100
$200
$150
Row
Max
Row
Min
Row
Ave
$100
$200
$300
3. Decision making under risk – the decision maker knows not
only the possible outcomes but also the probability of
occurrence for each outcome
 Choose the alternative with the largest MEAN payoff (EMV)
Demand 
Decision 
Order 10
Order 20
Order 30
States of Nature (Demand)
10
20
30
shirts
shirts
shirts
Prob(.3) Prob(.4) Prob(.3)
$100
$100
$100
$50
$200
$200
$0
$150
$300
EMV For
Decision
-- Expected value of perfect information (EVPI)
 EVPI = EMV under certainty - highest EMV under risk
 EMV under certainty =
(best outcome for SON1)x(prob. of SON1) +
(best outcome for SON2)x(prob. of SON2) +.....+
(best outcome for last SON)x(prob. of last SON)
Demand 
Decision 
10
20
30
shirts
shirts
shirts
Prob(.3) Prob(.4) Prob(.3)
$100
Order 10
$100
$100
$200
Order 20
$50
$200
$300
Order 30
$0
$150
EMV under certainty =
EMV For
Decision
$100
$155
$150
-- Decision tree - used to systematically represent problems that
involve sequential decision making
  represents a decision node, after which are all
alternatives the decision maker may choose
  represents a state of nature node, after which are all
outcomes (states of nature) may occur

Steps to represent a sequential decision problem by decision
tree
1) Define the problem
2) Structure (draw) the decision tree
3) Assign probabilities to each state of nature
4) Identify payoffs for each possible combination of
alternatives and states of nature
5) Compute the EMV for each state of nature node by working
backward
-- Example: Represent the T-shirt example by a decision tree
-- An advanced decision tree problem
Bakery Products is considering the introduction of a new line of
products. In order to produce the new line, the bakery is
considering either a major or minor renovation of the current
plant. The market for the new line of products could be either
favorable or unfavorable. Bakery Products has the option of not
developing the new product line at all. The following payoff
table has been developed for each alternative under various market
conditions.
Alternatives
Favorable
Unfavorable
Market
Market
Major Renovation
$100,000
-$90,000
Minor Renovation
$40,000
-$20,000
Do Nothing
$0
$0
The marketing department has estimated that the chance of having a
favorable market is about 60%.
1. Represent the problem by a decision tree
2. Which alternative maximizes the expected return (EMV)?
Additional information: Before making the final decision, Bakery
Products would like to consider a marketing research survey at a
cost of $5,000. Past experience indicates that the survey is
positive 80% of the time when the market is favorable and the
survey in negative 60% of the time when the market is unfavorable.
3. Redraw the decision tree to take the survey option into
consideration.
4. Should the company conduct the survey before making the final
decision? How should the decision be made if it is based on the
survey results (assuming now that the survey is done)?