Tversky and Kahnemann: Framing of Decisions

1
Tversky and Kahnemann Framing of Decisions

• Agenda
• Expected Utility Theory
• Violations of EUs axioms of choice Framing
  Effects
• Nontransparent Dominance
• Certainty and Pseudocertainty Effects
• Tradeoff Contrast
• Extremeness Aversion
• Conflict between descriptive and normative
  theories and conclusion

2
Expected Utility Theory

• Normative model of the modern theory of decision
  making under risk (Neumann Morgenstern)
• Four substantive assumptions
• 1. Cancellation
• elimination of any state of the world that yields
  the same outcome regardless of ones choice
• necessary for the representation of preference
  between prospects as the maximization of expected
  utility - (highly challenged and questioned)

3
Expected Utility Theory

• 2. Transitivity of preference
• necessary for the representation of preference by
  an ordinal utility scale u such that A is
  preferred to B whenever u(A) gt u(B) (independent
  on other available options - (questioned)
• 3. Dominance
• if one option is better than another in one state
  and at least as good in all other states, the
  dominant option should be chosen - (simple,
  compelling and accepted)
• 4. Invariance
• different representations of the same choice
  problem should yield the same preference,
  independent of their description - (basic
  principle, tacitly assumed, accepted)

4
Problem Violations of
axioms of choice

• Several experiments have proven that the
  deviations of actual behavior from the normative
  model are too widespread to be ignored, too
  systematic to be dismissed as random error, and
  too fundamental to be accommodated by relaxing
  the normative system (Tversky Kahnemann),
  e.g.
  Allais Paradox (1953) Ellsberg Paradox (1961)...
• Theory of rational decision cannot provide
  adequate description of choice behavior
• Kahnemann Tversky (1979) developed an
  alternative outcome-based approach called
  PROSPECT THEORY in order to explain many of the
  EU anomalies

5
Framing

• Expected utility theory assumes description
  invariance different formulations of the same
  choice problem should give rise to the same
  preference order
• Evidence that variations in the framing of
  options (in the description) yield systematically
  different preferences
• Framing effects lead to violation of the
  invariance and the dominance axioms of expected
  utility theory

6
Nontransparent Dominance

• Lottery Example Choose from a box with
  differently coloured marbles

Problem 1 (N88) Option A 90 white
6 red 1 green 1 blue
2 yellow 0 win 45
win 30 lose 15 lose
15 Option B 90 white 6 red
1 green 1 blue 2
yellow 0 win 45 win
45 lose 10 lose 15

0
100
Problem 2 (N124) Option C 90 white
6 red 1 green 3
yellow 0
win 45 win 30 lose
15 Option D 90 white 7 red
1 green 2 yellow
0 win 45 lose 10
lose 15
58
42
7
Nontransparent Dominance

• Results of Lottery Example
• Two formulations of the same problem elicit
  different preferences (violation of invariance)
• finer partition used in problem 1
• the number of positive and negative outcomes
  alone can enhance the attractiveness of an option
  (option C)
• Dominance rule is obeyed only when its
  application is transparent
• Dominance can be masked by a frame

8
Nontransparent Dominance

• Transparency in a Perceptual Example
  Müller-Lyer-Illusion
• whether relation of dominance is detected depends
  on
• Framing (transparency)
• Details of partitioning
• Sophistication and experience of decision maker

9
Certainty and Pseudocertainty Effects
Problem 1 (N77) Which of the following options
do you prefer? A. sure gain of 30 B. 80 chance
to win 45, 20 chance to win nothing
Problem 2 (N81) Which of the following options
do you prefer? C. 25 chance to win 30,
75 chance to win nothing D. 20 chance to
win 45, 80 chance to win nothing
10
Certainty Effect

• Problem 1 and 2 In problem 2, the probabilities
  of winning are reduced by a factor of four
• Preference switched due to certainty effect
  Reduction of probability of winning from
  certainty to 0.25 has greater effect than the
  reduction from 0.8 to 0.2

11
Certainty and Pseudocertainty Effects
Problem 1 (N77) Which of the following options
do you prefer? A. sure gain of 30 B. 80 chance
to win 45, 20 chance to win nothing
Problem 2 (N81) Which of the following options
do you prefer? C. 25 chance to win 30,
75 chance to win nothing D. 20 chance to
win 45, 80 chance to win nothing
Problem 3 (N85) Two-stage game Stage 1 75
chance to end the game without winning anything,
25 chance to move to stage 2 Stage 2 E. sure
win of 30 F. 80 chance to win 45, 20 chance
to win nothing
12
Pseudocertainty Effect

• Problem 2 and 3 Identical in terms of
  probabilities and outcomes
• Framing as two-stage game encourages the
  application of cancellation Failing to reach
  second step is discarded
• Pseudocertainty Effect Risk-averse choice in
  problem 3 but not in 2, outcome that is actually
  uncertain (3) is weighted as if it were certain

13
Certainty and Pseudocertainty Effects
Problem 1 Overweighing of the outcome that is
obtained with certainty due to loss aversion
Problem 2 Reduces probability of winning by
factor of 4
Problem 3 Introduces ex ante stage that gives
25 chance for second stage
Different choice due to certainty effect
violation of cancellation
Application of cancellation - same choice as in
problem 1 risk-averse choice due to
pseudocertainty
Same problem, differently framed
14
Framing - Findings

• Axioms of rational choice are
• generally satisfied in transparent situations
• often violated in nontransparent situations
• Framing enriches and complicates analysis of
  choice
• Framing of decisions depends on
• Language of presentation
• Nature of display
• Context of choice

15
Simonson and Tversky (1992)
Value Maximization

• Basic assumption of classical economic theory
  Consumer selects alternative with highest value,
  preference is independent of the context
• BUT consumer preferences are influenced by
  context of choice
• Effects of Context on Choice
• 1. TRADEOFF CONTRAST
• 2. EXTREMENESS AVERSION

16
1. Tradeoff Contrast

• People tend to compare an available option to
• other choices that are currently available (Local
  Effect)
• options that have been encountered in the past
  (Background Effect)
• e.g. a circle appears large when surrounded by
  small circles and small when surrounded by large
  ones
• Applies to single attributes (e.g. size) as well
  as to tradeoffs between attributes (e.g. price
  and quality)

17
Local Effect

• If y is clearly superior to z but x is not

• Tradeoff Contrast Hypothesis implies
• Addition of z can increase ys market share,
  yielding
• Violates value maximization principle popularity
  of an option cannot be increased by enlarging the
  offered set

? Asymmetric Dominance Effect
18
Enhancement Detraction

• Assumption no strong preference between x and z
  z is dominated by y but not by x
• Tradeoff Contrast Hypothesis offered set will
  affect choice even when no option has a decisive
  advantage over another

Once a third option is introduced, the decision
maker can compare three tradeoffs, which...

• Enhances the attractiveness of y y fares better
  in triple than in pairs (explained by extremeness
  aversion)

• Detracts from the attractiveness of w w fares
  worse in triple than in pairs

19
2. Extremeness Aversion (1/3)

• Implications of value maximization
• If y is between x and z on all relevant
  attributes, then
• Middle option (y) is expected to lose relatively
  more than the other extreme option x from the
  introduction of z
• At variance with extremeness aversion y will
  lose relatively less than x from the introduction
  of z

20
Extremeness Aversion (2/3)

• Main features
• An option is more attractive to the respondent if
  it is an intermediate option in a choice set
• Attractiveness is lower for extreme options
• Based on Principle of Loss Aversion losses loom
  larger than gains
• Alternatives are evaluated in terms of their
  advantages and disadvantages relative to other
  options disadvantages are weighted more heavily
  than advantages

21
Extremeness Aversion (3/3)

• Each extreme option (x and z) has a large
  advantage and a large disadvantage relative to
  the other extreme small advantage and small
  disadvantage relative to the middle option (y)
• Middle option (y) has small advantages and small
  disadvantages relative to both extremes
• If (pairwise) disadvantages loom larger than the
  corresponding advantages, the middle option
  should fare better in the triple than in the
  pairs.

22
Example for Extremeness Aversion

• In group 1 2, the cameras varied across three
  dimensions
• two qualitative dimensions (optics, resolution)
• one price dimension
• Group 3 introduction of a forth dimension
  (recording of video sequences) for camera 3 in
  order to emphasize its quality and innovativeness

23
Theory

• Framing effects and associated failures of
  invariance are ubiquitous - cant be ignored by
  adequate descriptive theory

• Invariance is normatively indispensable - no
  adequate prescriptive theory should permit its
  violation

• Seems unrealizable to construct a theory that is
  both descriptively and normatively acceptable

24
Prospect Theory

• Several models of risky choice tried to explain
  the observed violations of expected utility
• PROSPECT THEORY differs by being descriptive,
  without normative claims
• Explains preferences, whether or not they can be
  rationalized
• Accommodates violation of dominance and
  invariance, especially in nontransparent
  situations

25
Conclusion

• To retain rational model in its descriptive role,
  relevant bolstering assumptions must be validated
• Defence of rationality The substantial
  violations of the standard model are
• restricted to insignificant choice problems
• attributable to the cost of thinking and will
  thus be eliminated by proper incentives
• quickly eliminated by learning (accurate and
  immediate feedback)
• inconsequential because of corrective effects of
  the market
• Where they fail, implications of the descriptive
  analysis must be traced

26
Questions?

• Thanks for your attention.

27
Nontransparent Dominance

• Transparency in a Perceptual Example
  Müller-Lyer-Illusion
• whether relation of dominance is detected depends
  on
• Framing (transparency)
• Details of partitioning
• Sophistication and experience of decision maker

28
Compromise and Polarization

• Both inequalities hold
• Addition of an extreme option increases share of
  the middle option relative to the other extreme
• Expected if disadvantages loom larger than
  advantages on both attributes

• Only one inequality holds (disadvantages loom
  larger than advantages only on one dimension but
  not on the other)
• e.g. extremeness aversion for quality, little or
  no extremeness aversion for price
• Consumers find lowest quality more aversive than
  highest price
• Asymmetry between price and quality