Upload
aron-barrett
View
217
Download
0
Embed Size (px)
Citation preview
Representativeness Heuristic
Then: Framing Effects
Psychology 355: Cognitive Psychology
Instructor: John Miyamoto
6/2/2015: Lecture 10-2
This Powerpoint presentation may contain macros that were used to create the slides. The macros aren’t needed to view the slides. If necessary, you can disable the macros without any change to the presentation.
Outline
• Finish discussion of the availability heuristic
• Representativeness heuristic
• Preferential choiceFraming effects
Psych 355, Miyamoto, Spr '15 2
Lecture probably ends here
Main Claims of the Heuristics & Biases Movement
3Psych 355, Miyamoto, Spr '15
Main Claims of the Heuristics & Biases (H&B) Movement
• Human cognitive processes do not follow the pattern
of a rational model. o (Rational model = expected utility theory & Bayesian decision model)
• Human decision making uses heuristic reasoning strategies –
reasoning strategies that are useful because they are easy and
generally effective, even though they can lead to errors.
Heuristic reasoning strategies ....o .... are often fast and effective, o .... place low demands on cognitive resources.o .... but they can lead to errors in particular situations.
List of Many Heuristics that are Studied in the Heuristics & Biases Program
4Psych 355, Miyamoto, Spr '15
Some Heuristics in Inductive Reasoning
• Availability
• Representativeness
• Anchoring & Adjustment
• Confirmation bias
• Focusing illusion
• Framing effects
• Mental accounting
• More heuristics have been proposed than are listed here.
Availability Heuristic
5Psych 355, Miyamoto, Spr '15
“Other Factors” that Influence the Availability of Events
Continuation of this Slide: List Examples of Other Factors
6Psych 355, Miyamoto, Spr '15
“Other Factors” that Influence the Availability of Events
Examples of
“Other Factors”
• Egocentric bias.
• Dramatic events seem more common than non-dramatic events.
• Recent events seem more common than earlier events.
• Biases in the media create biases in the availability of stereotypes.
Effects of Information Sampling Bias in the Media
7Psych 355, Miyamoto, Spr '15
Information Sampling Bias in Everyday Media
Things we all know:o TV ads do not give an accurate picture of the value of products.o Political spin doctors are trying to manipulate our beliefs.o TV news is emphasizes dramatic events; it ignores undramatic events.o The portrayal of men/women, black/whites, rich/poor, gay/straight,
on TV is not a representative presentation of these groups.o Our own experiences are not typical of everybody’s experience.o Etc.
• We all know that these information sources are biased, but can
we really correct for these biases when forming beliefs? Doubtful.
Summary re Availability Heuristic - Skip Over
{Biases∈¿ InformationSources } { Biases∈¿ Availability } {Biases∈¿ Perceived Likelihood
of Events }
Summary re the Availability Heuristic
• Judging probability in terms of availability is a heuristic.♦ I.e., it is generally a reasonable way to estimate likelihood, but it can lead
to certain systematic errors.
• Factors that are not related to experienced frequency can makemake particular events very available.
♦ E.g., the perceived probability of being killed by a random crazy person will tend to be exaggerated.
Psych 355, Miyamoto, Spr '15 8Introduction to the Representativeness Heuristic
9Psych 355, Miyamoto, Spr '15
Representativeness Heuristic
Representativeness Heuristic:
Events that are more representative are regarded as more probable.
"more representative"
means
"more similar to a stereotype
or to a typical member of a class."
Event A is more representative
than Event B
Event A is more probable
than Event B
Example of Jim: An Athletic, Muscular & Competitive Guy
Psych 355, Miyamoto, Spr '15 10
Question: Jim is tall and very muscular. He's also very competitive.
He drives an expensive car and wears flashy clothing.
Which is more probable?
a) Jim is a professional athlete.
b) Jim is a lawyer or financial analyst.
• People predict that Jim is a professional athlete because
Jim is similar to a stereotype of a professional athlete.
• It is a better bet that Jim is a lawyer or financial analyst
because there are many more lawyers and financial analysts than
professional athletes.
This response is predicted by the Representativeness Heuristic
Representativeness Heuristic – An Example
Return to Slide with Diagram of Representativeness Heuristic
This is the better bet.
11
Representativeness Heuristic:
Events that are more representative are regarded as more probable.
o Example: Jim is muscular/athletic/competitive.
Psych 355, Miyamoto, Spr '15
Jim is more similar to the stereotype.
Jim is less similar to the stereotype.
Representativeness Heuristic
Event A is more representative
than Event B
Event A is more probable
than Event B
a professional athlete?
a lawyer or financial analyst?Is he .....
Intro to the Lawyer/Engineer Problem
12Psych 355, Miyamoto, Spr '15
Lawyer/Engineer Problem (K&T, 1973)
DESCRIPTION OF JACK: Jack is a 45-year-old man. He is married and has four children. He is generally conservative,
careful, and ambitious. He shows no interest in political and social issues.
• 30:70 Condition: High Base Rate for Engineer
If Jack's description were drawn at random from a set of
30 descriptions of lawyers and 70 descriptions of engineers,
what would be the probability that Jack is one of the engineers?
• 70:30 Condition: Low Base Rate for Engineer
If Jack's description were drawn at random from a set of
70 descriptions of lawyers and 30 descriptions of engineers,
what would be the probability that Jack is one of the engineers?
Findings re Lawyer/Engineer Problem
13Psych 355, Miyamoto, Spr '15
Results re Lawyer/Engineer Problem
• Probability of "engineer" was rated to be about the same
in the low and high base rate conditions.
(Insensitivity to Base Rate)o High base rate condition = 30:70 Condition
Low base rate condition = 70:30 Conditiono Probability theory implies that Jack is much more likely to be
an engineer in the high base rate condition than in the low base rate condition.
• Why do people ignore base rates? See next slide
Why Do People Ignore Base Rates? The Representativeness Explanation
14Psych 355, Miyamoto, Spr '15
Why Do People Often Ignore Base Rates?
The Representativeness Heuristic: People judge probability based
on the similarity of the current case to a stereotype.
(a) Jack is equally similar to a typical engineer in the
low and high base rate conditions.
(b) People ignore the base rate because the base rate is irrelevant to
the judgment of how similar Jack is to a typical engineer.
• Probability theory shows that the base rate is very relevant
to judging the probability that Jack is an engineer.
• Cognitive theory shows that the base rate is often not
psychologically relevant to judging the probability that Jack is an
engineer.
When Does It Matter Whether People Ignore Base Rates?
15Psych 355, Miyamoto, Spr '15
When Does It Matter Whether People Ignore Base Rates?
• Evidence shows that physicians sometimes overlook base rates
when attempting to diagnose a disease.
• Evidence suggests that investors are overly influenced by
short-term information regarding the value of stocks.
• Business decisions tend to be overly influenced by short-term
trends.
Criticism of Goldstein’s Description of the Lawyer/Engineer Problem
16Psych 355, Miyamoto, Spr '15
Criticism of Goldstein’s Description of the Lawyer/Engineer Problem
• The Goldstein description of this study is inadequate because it does
not contrast the 30:70 condition with the 70:30 condition. It only
mentions the 70:30 condition.
• The important finding is that subjects in the 30:70 and 70:30
conditions are equally confident that Jack is an engineer (subjects in
the two conditions overlook the difference in the base rate). o Knowing only the result for the 70:30 condition does not establish that subjects
ignore base rates.o See Goldstein p. 374.
Ignorance of Regression Effects
Statistical Theory Implies that Regression Effects Will Occur
• Very tall fathers tend to have sons who are not quite as tall.
• Very short fathers tend to have sons who are somewhat taller.
• Why?
Psych 355, Miyamoto, Spr '15 17
very tall father
fairly tall son
very short father
fairly short son
Statistical Theory Implies that Regression Effects Will Occur
• Statistical reason for regression effects: A predicted value will be closer to the mean than is the variable on which the prediction is based.
Zpredicted Y = ZX
Zpredicted Y = predicted z-score for Y
= the population correlation between X and Y
ZX = z-score for the predictor X
• Implication: If X and Y are not perfectly correlated, then the predicted value of Y is always closer to its mean than the value of X.
Psych 355, Miyamoto, Spr '15 18
very tall father
fairly tall son
very short father
fairly short son
Psych 355, Miyamoto, Spr '15 19
People's Tendency to Ignore Regression Effects
• Sophomore Slump: A baseball player who does exceptionally well
during his rookie season often does noticeably worse during his
sophomore (second) season. Why does this happen?
• Reason: A player’s batting average during the rookie season is not
perfectly correlated with his batting average during the sophomore
season. Therefore, on the average, a regression effect is inevitable.o Same holds for any other statistic, like number of home runs, stolen bases,
earned run average (for a pitcher), etc.
Misconceptions of Regressions – Other Examples
Psych 355, Miyamoto, Spr '15 20
Misconceptions of Regression – Other Examples
• Israeli flight instructors and the effects of praise and punishment.
• Evaluating medical treatments or psychotherapies that select
patients who are already in extreme difficulty.
Why Does the Representativeness Heuristic Cause People to Overlook Regression Effects?
Why Does the Representativeness Heuristic Cause People to Overlook Regression Effects?
• Suppose a pilot just made a terrible landing. What should you predict for the next landing?
♦ Another terrible landing (most similar outcome)
A bad landing that is not as bad as the first landing (less similar to the previous landing, but it is more probable because of regression to the mean).
• People predict another terrible landing because it is the most similar outcome, while ignoring a factor, regression to the mean, that is statistically relevant, but not related to similarity.
Psych 355, Miyamoto, Spr '15 21Conjunction Fallacies
22Psych 355, Miyamoto, Spr '15
Conjunction Fallacies – The Famous Linda Problem
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 anti-nuclear demonstrations.
F: Judge the probability that Linda is a feminist.
T: Judge the probability that Linda is a bank teller.
F & T: Judge the probability that Linda is a feminist
and a bank teller.
• Probability Theory: P(F) ≥ P(F & T), P(T) ≥ P(F & T)
• Typical Judgment: P(F) > P(F & T) > P(T)
Why Are Conjunction Fallacies Psychologically Interesting?
23Psych 355, Miyamoto, Spr '15
Why Conjunction Fallacies Are Psychologically Interesting?
• Conjunction fallacies strongly support the claim:
Human reasoning with uncertainty is different from
probability theory.o Human reasoning with uncertainty is based on a various heuristics – the
conjunction fallacy is caused by the use of a representativeness heuristic.
Two Question Regarding Conjunction Fallacies:
• What is wrong with the judgment pattern:
P(F) > P(F & T) > P(T)?
• Why do people's judgments have this pattern?
Probability & the Set Inclusion Principle
24Psych 355, Miyamoto, Spr '15
Probability and the Set Inclusion Principle
• If set B is a subset of set A, then
the probability of B must be
equal or less than the probability
of A.
B A P(B) < P(A)
Rationale: When B occurs,
A also occurs, so the probability
of B cannot exceed the
probability of A.
A B
Sample Space (set of all possibilities)
Interpretation of Linda Problem in terms of Set Inclusion
25Psych 355, Miyamoto, Spr '15
F: Judge the probability that Linda is a feminist.
T: Judge the probability that Linda is a bank teller.
F & T:Judge the probability that Linda is a feminist and a bank teller.
• Probability Theory: P(F) ≥ P(F & T), P(T) ≥ P(F & T)
• Typical Judgment: P(F) > P(F & T) > P(T)
Why Do People Make Conjunction Errors?
Conjunction Fallacy
Sample Space
F F & T T
Linda Problem: 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 anti-nuclear demonstrations.
Psych 355, Miyamoto, Spr '15 26
Tuesday, June 02, 2015: The Lecture Ended Here
27Psych 355, Miyamoto, Spr '15
Why Do People Make Conjunction Errors?
• Remember: The representativeness heuristic predicts that people judge the
probability based on how similar the individual case is to a typical member
(stereotype) of a group.
• The description of Linda sounds more similar to someone who is a feminist
and a bank teller, than to someone who is only a bank teller.
Criticisms of the Representativeness Explanation of Conjunction Fallacies
stronger similarity
Descriptionof Linda
Bank TellerPrototype
Feminist Bank TellerPrototype
weaker similarity
28Psych 355, Miyamoto, Spr '15
Criticisms of This Interpretation
• Criticism: The Linda problem is just one problem.
Reply: Same pattern is found with many similar problems.
• Criticism: Maybe people think “bank teller” means someone who is
a bank teller and not a feminist.
• Criticism: Conjunction errors can be eliminated by stating
the question in terms of frequencies instead of probabilities.
Summary re Representativeness Heuristic
29Psych 355, Miyamoto, Spr '15
Summary re Representativeness Heuristic
• There is nothing wrong with using similarity as a relevant factor
in judging a probability. o The problem is that attention to similarity causes people to ignore
other factors, like base rates, regression effects and set inclusion, that are also relevant to judging probability.
Consequences of the Use of the Representativeness Heuristic
• Base rate neglect
• Conjunction errors
• Overlooking the importance of sample size
• Overlooking regression effects
Two Major Issues in Psych of Decision Making - Probability & Preference