Day 2
Evolution of Decision-Making


Tversky and Kahneman, 1974
Heuristics – general rules of thumb, or habits
 Generally result in decent estimates
 Can be fooled with systematic biases
2

Judging probabilities by the degree to which A is
representative of B

Linda is 31 years old, single, outspoken, and very
bright. She majored in philosophy. As a student,
she was deeply concerned with issues of
discrimination and social justice, and also
participated in antinuclear demonstrations.
Please check the most likely alternative:
 Linda is a bank teller
 Linda is a bank teller and is active in the feminist
movement
3


Most people (9 out of 10) answer B
However, B is more specific than A and is a
smaller subset of the population, therefore is
not more likely
Bank Tellers
Feminists
Bank Tellers
who are also
Feminists
4


Belief that random samples of a population
will resemble each other
The mean IQ of a population of eighth
graders in a city is known to be 100. You have
selected a random sample of 50 children for a
study of educational achievements. The first
child tested has an IQ of 150. What do you
expect the mean IQ to be for the whole
sample?
5






First child has IQ of 150
Remaining 40 have mean IQ of 100
Total is 5050, average is 101, not 100
We expect remaining sample to somehow
“balance out”
Small samples don’t randomly cancel out
outliers with other outlier values
Fallacy of the Hot Hand
6



Heuristic in which decision-makers “assess
the frequency of a class or the probability of
an event by the ease with which instances or
occurrences can be brought to mind
What is a more likely cause of death in the US
– being killed by falling airplane parts or by a
shark?
Death by falling airplane parts is 30 times
more likely, but shark death is more easily
imagined
7


When imagination is limited
When imagining an event is so upsetting that
it leads to denial
8
9


Tversky and Kahneman 1981 define framing
as “the decision maker’s conception of the
acts, outcomes, and contingencies associated
with a particular choice”
Frames are partly controlled by formulation
of the problem, and partly controlled by
norms, habits, and characteristics of the
decision maker
10

Decision 1 (risk aversion with gains at stake)
 Alternative A – a sure gain of $240 Preferred
 Alternative B – a 25% chance to gain $1000, and a
75% chance to gain nothing

Decision 2 (risk seeking with losses at stake)
 A sure loss of $750
 A 75% chance to lose $1000, and a 25% chance to
lose nothing
Preferred
11



Classic exercise in loss aversion
Two choices presented to respondents
They had to choose option A or option B
12

The United States is preparing for the
outbreak of an unusual Asian disease, which
is expected to kill six hundred people. If
program A is adopted, two hundred people
will be saved; if program B is adopted, there
is a one-third probability that six hundred
people will be saved and a two-thirds
probability that no people will be saved.
Which program do you favor?
13

When asked of physicians, 72% chose option
A, the safe-and-sure strategy
14

The United States is preparing for the
outbreak of an unusual Asian disease, which
is expected to kill six hundred people. If
program C is adopted, four hundred people
will die. If program D is adopted, there is a
one-third probability that nobody will die and
a two-thirds probability that six hundred will
die. Which of the two programs do you
favor?
15



When described in terms of deaths rather
than lives saved, physicians reversed their
choices, with 78% selecting option D, the
risky strategy
Both scenarios are identical in lives lost or
saved
Loss aversion is a way of skipping the math
and using emotion to make the decision
16
17



Bernoulli 1713
Linear model of utilities
Six axioms of expected utility theory Plous p. 81
 Ordering of alternatives – comparing choices
 Dominance – some better than others
 Cancellation – common factors cancel out
 Transitivity – if a>b and b>c, then a>c
 Continuity – gamble preferred over intermediate
 Invariance – not affected by presentation style
18






Herbert Simon, 1955
First major advance in decision theory since
Bernoulli’s in 1700’s
Better match to real world decision-making
Satisfice rather than optimize
Satisficing finds alternative that meets most
of the major criteria, then stops
Example – apartment search in Arlington
19
Feature 1 Feature 2
Feature 3
Feature 4…
Feature n
Price
Nearness
Size
Security
Neighbors
Complex 1
$650
2 miles
800 sq ft
Limited
Flight att.
Complex 2
$600
3 miles
810 sq ft
None
Cars on blocks
Complex 3
$350
4 miles
600 sq ft
None
Lots of kids
Complex 4
$850
1 mile
900 sq ft
Gated
Nearby sorority
Complex 5… $1200
Complex n
$900
20


Lichtenstein (1971)
Utility theory doesn’t quite all aspects of
consumer behavior
 Rating of attractiveness (a weight applied to a
probably outcome)
 A gamble is seen by the authors as a multidimensional stimulus

Some dimensions of affect are playing a role
in what should be a cognitive decision
21


Kahneman and Tversky, 1979
Replaces “utility” concept with “value”
 Utility is defined in terms of net worth
 Value is defined in terms of gains and losses


Losses loom larger than gains
Endowment effect
 What one owns is more valuable than what
someone else owns
22



The value of each outcome is multiplied by a
decision weight
Decision weights are very subject to biases
This leads to a decidedly n0n-symmetrical
value function, where the value function for
losses is decidedly steeper than that for gains
23


Outcomes with small probabilities are
overweighted relative to outcomes with
higher certainty
This tendency leads to the concepts of
insurance and gambling as industries
24
High Value
Losses
Gains
Low Value
25




Example of $1 found by a millionaire or a
homeless person – who values the
incremental $1 more?
Who would be more concerned with the loss
of that incremental $1?
For gains, this produces risk aversion
For losses, this produces gambling
26

Damasio and Loewenstein investing game
 In each round, subject decides to invest $1 or
invest nothing
 No invest, subject keeps dollar
 Invest, researcher flips coin for $1 loss or $2.50
gain
 Rational investors should always choose to invest
27



Decisions made relative to a reference point
Comparison of imaginary outcomes referred
to as “counterfactutal reasoning”
Regret is based on two assumptions:
 People experience sensations of regret and
rejoicing
 When making decision under uncertainty, people
try to anticipate and take into account these
sensations
28
29

People tend to evaluate personal risk
outcomes based on the valence
 Positive outcomes – more probable
 Negative outcomes – less probable
 Rose colored glasses as a lens to our lives
30




Conjunctive compound events are A+B,
where A and B are simple events
Compound events are preferred when
conjunctive
Simple events are preferred when compound
events are disjunctive (A or B)
People anchor on the probabilities of the
simple events that make up the compound
event and fail to adjust probabilities
31


Once we estimate probabilities, we are slow
to modify those estimates
When modified, the estimates are changed
more slowly than the data would dictate
32

Three dimensions for public perception of risk
(Slovic, 1987)
 Dread risk – lack of control, catastrophic potential
 Unknown risk – risks that are unobservable
 Magnitude of risk – number of people exposed to
it
33



Maintain accurate records – minimize
primacy and recency effects
Beware of wishful thinking – wishing for
positive outcomes
Disaggregate compound events into simple
events
34

What is correlation?

What is causation?

Are they the same?
35

What is correlation? Degree of covariation
between two variables

What is causation? The outcome of one
event resulting in the outcome of another
event

Are they the same? No – a common mistake
made by many market researchers
36


Illusory - People can “see” a correlation
between objects based on the objects
semantic similarities, even when no
correlation exists
Invisible – People fail to see a correlation even
when it exists – our expectations of visible
patterns causes us to miss some strong but
unexpected patterns
37




Einhorn and Hogarth, 1986
Correlation need not imply causal connection
Causation need not imply a strong correlation
Some people believe that causation implies
correlation – they called it “causalation”
38


How people make causal attributions (Kelley,
1967)
Three main variables to explain behavior
 The person
 The entity – feature of the situation
 The time – feature of the occasion

Based on three sources of information
 Consensus, distinctiveness, and consistency
39



Sometimes people ignore base rate
information
Sometimes people focus on salient, available,
or vivid information
Fundamental attribution error (Ross, 1958) is
that people’s behaviors tend to swamp all
other situational variables
40




When faced with a successful outcome,
people are more likely to accept responsibility
and take more credit for the outcome
When faced with an unsuccessful outcome,
people are more likely to attribute blame to
others
Ego-centric biases – married couple example
Positivity effect – attribute positive behaviors
to dispositional factors and negative
behaviors to situational factors
41