Descriptive Decision Making Models

1
Descriptive Decision Making Models

• how people ACTUALLY make decisions

2
Herbert Simon - satisficing

• Why do people
• SATISFICE
• rather than
• OPTIMIZE?

3
Prospect Theory ( Kahneman
Tverskey)

• VALUE
• value function
• Gains
• 500
• Losses
• loom larger
• -500

4
Prospect theory

• GAINS loss aversion
• incumbent politician

• LOSSES felt more strongly than gains
• endowment
• effect

5
Cognitive basis for STATUS QUO bias?

• PUBLIC POLICY
• LEGAL rulings

6
Framing effects

• Importance of
• reference effect
• outcome as gain
• .risk adversive
• outcome as lossrisk seeking

7
Preferences

• Decision WEIGHTS
• not
• Probabilities
• .overweight
• small Probabilities

• Which prefer?
• A 1/1000 chance winning 5000
• or
• B A sure gain of
• 5
• 3/4 prefer risky choice

8
Preferences cont..

• Which prefer?
• C A 1/1000 chance losing 5000
• D A sure loss of 5
• (4/5 a sure loss)

9
Other differences - SEU

• Certainty Effect

• Would you buy insurance with half premium of
  regular but only payoff 50 of time?
• (preference to eliminate risk not reduce it)

10
Other differences - SEU

• Pseudo-certainty

• Vaccine-50 population from disease expected to
  infect 20
• OR
• Complete protection from disease infect 10 but
  no protection other strain

11
3rd Model Regret Theory

• ASSUMES
• under UNCERTAINTY
• People anticipate and take into account REGRET
  
• REJOICING

12
Example Regret theory

• Which would you choose?
• 1000 for sure
• OR

• 2000 if coin comes up heads
• avoid regret

13
Who would you vote for?

• INCUMBENT REPRESENTATIVE in Congress
• OR
• UNKNOWN (inexperienced) challenger

14
Other Models Multi-attribute Model

• Approximates a linear model
• Simple regression can approximate
• College admission board
• (outperform)
• GPA SAT test scores rec letters

15
Multi-attribute Model

• Factors approximating College admission board
• GPA SAT test scores rec
• letters

16
Additive with trade-offs

• Add differences
• Weight differences and sum
• Example focus on key difference 2
  alternatives (cars)
• Ideal point model (ideal PREZ)
  

17
People often DONt make trade-offs

• eliminate any outside cut-off (noncompensatory)
• evaluate on basis attribute (disjunctive rule)
• evaluate on most important dimension
  (lexicographic)
• elimination-by-aspects (automatic,
• price, color.N1)

18
What if 2 equally valued alternatives?

• What do people do?
• (select one superior on most important dimension)
• its a cool car!

19
People are binary choosers

• Why
• most important
• hypothesis
• so often true
• (Slovic)

• I.e., Choose candidate X
• strongest Pro-Choice candidate
• (satificing on important dimension)

20
So what will people do when forced to make a
decision

• Under uncertainty?
• Incomplete information
• SIMPLIFY on the basis of some rule
• take the job that
• pays the most

21
Implication

• Hard to predict in advance what the
• specific rule.
• choice will be
• WHY?

22
Political problems ill-structured problems

• No Simple choice
• Not single person
• Small Group

• STEPS
• 1 develop shared problem representation
• 2 discuss-reach consensus
• 3 Development problem solution
• (key reference point
• STATUS QUO - make a minor adjustment

23
Develop a shared problem representation
(immediate problem at hand

• Define Problem
• .
• .
• Discuss (what are we doing now?)
• .
• .
• shared solution

• CUBAN MISSILE
• CRISIS
• Must get the missiles out before next election
• .
• .
• blockade use force if necessary