# MGTC20 Final.docx

39 views21 pages
12 Apr 2012
School
Course
1
SECTION 4: HEURISTICS & BIASES
CHAPTER 10 THE REPRESENTATIVENESS HEURISTIC
Heuristics General rules of thumb.
Shortcuts can yield very close approximation to the “optimal” answers, but can also lead to predictable
biases & inconsistencies
Will focus on 2 related issues:
o process by which decision makers reach their conclusions
o the biases that can result as a consequence of these processes
Advantages- Reduce time/effort required to make reasonably good judgments/decisions
The ABC/s of Representativeness
Ppl often judge probabilities by the degree to which A is representative of B (Degree that A resembles B)
o Called the “Irepresentativeness Heuristic
o Eg A is a person, B is a group (if trying to estimate prob that A came from B)
o Eg A is an event/effect (six heads in a row), B is process/cause (flipping coin)
Textbook example Linda is concerned w/ issues of discrimination & social justice
o Alternatives: 1. Bank teller. 2. Bank teller & active in feminist movement
o 90% believed she was a bank teller & active in the feminist movement.
o Most ppl believe the more specific event is more probable than a general event
o Violates fundamental rule of probability:
Conjunction Fallacy - Conjunction/Co-occurrence of 2 events CANNOT be more likely than the
prob of either event alone.
Tversky & Kahneman concluded as the amt of detail in the scenario increases, its prob can only decrease
steadily, but it representativeness & hence it’s apparent likelihood increase
The reliance on representativeness is a primary reason for the unwarranted appeal of detailed scenarios &
the illusionary sense of insight that constructions provide
Spec scenarios appear more likely b/c they are more representative of how we imagine particular events
The Law of Small Numbers:
The Law of Small Numbers Belief that random samples of a population will resemble each other & the
population more closely than statistical sampling theory would predict
o Reference to a law in stats known as the law of large numbers (larger a sample you draw from the
population, the closer its average will be to the population average)
Eg If ppl are asked to write down a random sequence of coin tosses w/o flipping a coin, they often try to
make the string look random at every point (called “Local representativeness)
Eg mean IQ = 100. First child (of 50) tested has IQ of 150. What do you expect mean IQ to be?
o Most ppl will say 100, bc they think there will be low IQ scores to “balance out” the high score of 150.
This assumes that chance is self-correcting. However, chance is not self correcting it does not cancel
out/correct high scores. It merely “dilutes” high scores w/ other scores that are closer to the avg
Tendency to view chance as self-correcting is an example of a bias resulting from the representativeness
heuristic (bc samples are expected to be highly representative of their parent population)
Gambler’s fallacy belief that a successful outcome is due after a run of bad luck (belief that independent
trials w/ the same outcome will soon be followed by an opposite outcome) .. also caused by representative
heuristic
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 21 pages and 3 million more documents.

2
The Hot Hand
Demonstration of the law of small numbers
Hot Hand (aka “streak shooter”) player who has a better chance of making a basket after one or more
successful shots, than after missing a shot
Truth is chances of making next basket were not significantly diff from the player’s overall probability of
Experiment Subjects viewed six diff series of X’s & O’s. Each series contained 11 X’s and 10 O’s & the prob
of alternating b/w the two letters was set from 0.4 0.9
o Subjects selected the 0.7 & 0.8 sequences as the best examples of a chance series, rather than
selecting the 0.5 sequence
o 62% classified the 0.5 sequence as “streak shooting”
Neglecting Base Rates
A reliance on representativeness can lead ppl to ignore “base rate” info (relative freq w/ which an event
occurs)
When asked to rate the chance of a randomly selected person being an engineer, they used the base rate
(30%)
When given descriptive info (even uninformative info) they tended to ignore the base rates (50% if no
concrete info was given)
Ppl often use base rate info when it is consistent w/ their intuitive theories of cause & effect (eg. Number of
hours studying/ week vs weekly income)
Nonregressive Prediction
Ppl tend to neglect the diagnosticity of the info on which they base their prediction, & as a result make
“nonregressive” predictions
Regression to the mean stat phenomenon in which high/low scores tend to be followed by more avg
scores
Eg Scores on test are moderately related to GPA. What GPA would you predict for student who scored
725?
o Most students predict 3.5 3.7. The best prediction lies b/w 2.5(avg) 3.6
Most psychologists think of test scores as being made up of two independent components:
o True Score what student would score if test were perfect measure of ability
o Error result of all factors that have nothing to do with ability (sleep, light,etc..)
In most cases these factors tend to cancel each other out. Occasionally they combine to dramatically
increase or decrease a test score. In future, test scores are likely to regress toward the true score
Tendency to overlook regression can lead to critical errors in judgment
o Mislabeling of simple regression phenomena (extremely good/bad performances followed by less
extreme performances) eg “sports illustrated jinx”
Clinical vs Actuarial Prediction
Accuracy of “actuarial” predictions (based solely on empirical relations b/w a given set of variables & an
outcome) is equal to or better than the accuracy of “clinical” predictions (based on the judgment of human
beings)
o Predictions are more accurate when they are not made by a human decision maker
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 21 pages and 3 million more documents.

3
Conclusion
Several ways to improve judgment & decision making skills:
o Don’t be misled by highly detailed scenarios
o Whenever possible, pay attention to base rates (particularly imp. when event is rare/uncommon)
o Remember chance is not self correcting (past evens do not effect future outcome)
o Don’t misinterpret regression toward the mean
CHAPTER 11 THE AVALABILITY HEURISTIC
Availability heuristic rule of thumb in which decision makers assess the freq of a class or the prob of an
event by the ease w/ which instances/occurrences can be brought to the mind
o Usually works well.. common events easier to remember/imagine than uncommon. Can estimate
freq/probability & can simplify otherwise diff judgments
This chapter examines 3 common questions:
o What are the instances in which the availability heuristic leads to biased judgments?
o Do decision makers perceive an event as more likely after they have imagined it?
o How is vivid info diff from other info?
Availability Goes Awry
Most ppl rate shark attacks as more probable than death from falling airplane parts b/c shark attacks
o Yet chances of dying are 30X greater by falling airplane parts
o Availability is misleading indicator of freq
o See textbook example are there more words that begin with K or that K as the 3rd letter? are there
more paths in A or B?
An Imaginative Study
John Carroll published study that linked the availability heuristic w/ the act of imagining an event.
o If easily imagined events are judged to be probable, then the very act of imagining an event will
increase its availability and make it appear more likely
o Experiment Subjects asked to imagine watching the televised coverage of the election either the
night of or the following morn. Half were told Ford wins & wins much of the Midwest & West, etc..
Other half told to imagine Carter wins due to strength in South, etc..
Next, asked to predict actual outcome. Subjects predicted that the subject they imagined winning
would actually win.
The Limits of Imagination
What if outcome is difficult to imagine?.. Does the perceived likelihood od that outcome increase of
decrease?
In difficult to imagine condition (virus on campus), subjects read about disease with abstract symptoms
(vague sense of disorientation, malfunctioning nervous symptoms, etc..)
o Control group asked how likely they were to contract it (easy to imagine or difficult to imagine)
o Experimental group read about disease & imagine a 3-week period during which they contracted the
symptoms & write detailed descriptions of how they felt during that time
Findings Control subjects were not significantly influenced by how easy symptoms were to imagine, but
experimental subjects were strongly affected
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 21 pages and 3 million more documents.