Heuristics & Biases
date post
25-Feb-2016Category
Documents
view
50download
4
Embed Size (px)
description
Transcript of Heuristics & Biases
Heuristics & BiasesMAR 3053
February 28, 2012
Part 1: Heuristics & intuitive judgmentThe use and misuse of affect, availability, representative-ness, and anchors
Two systems of reasoningSystem 1IntuitiveAutomaticEffortlessRapid & parallelConcreteAssociativeSystem 2ReflectiveControlledEffortfulSlow & often serialMay be abstractRule-based
Which bet would you choose?1 in 109 in 100
Who chooses the large box?Percentage of participants choosing the box with greater # of total balls(odds with small box = 10%; odds with large box = value shown on x-axis)
Chart1
60
55
41
30
24
Series 1
Sheet1
Series 1Series 2Series 3
9%602.42
8%554.42
7%411.83
6%302.85
5%24
To resize chart data range, drag lower right corner of range.
What is a heuristic?Mental shortcut used in judgment and decision makingEssential for living in an uncertain worldBut they can lead to faulty beliefs and suboptimal decisionsBy looking at errors and biases, we can learn how people are reasoning under uncertainty
Two types of heuristicsSpecial purpose heuristics use restricted to specific domainsHeight as a guide for ability as basketball player# of publications as guide for quality as an academicGeneral use heuristicsAffectAvailabilityRepresentativeness (similarity)
The affect heuristic## migrating birds die each year by drowning in uncovered oil ponds, which the birds mistake for bodies of water. Covering the ponds with nets could prevent these deaths. How much money would you be willing to pay to provide the needed nets?2,000 birds -- $8020,000 birds -- $78200,000 birds -- $88
The identifiable victim effectA death of a single Russian solder is a tragedy. A million deaths is a statistic. Joseph Stalin
AffectJudgments of life happiness:People asked 2 questions:1) How satisfied are you with your life these days?2) How many dates have you had in the last month?Correlation = -.12Another group asked in opposite order 2), then 1)Correlation = .66Strack et al., 1993
The availability heuristicMaking judgments about the frequency or likelihood of an event based on the ease with which evidence or examples come to mindExample: Category size
Kansas?Nebraska?
AvailabilityEgocentric allocations of responsibility: OverclaimingPeople claim more responsibility for collective endeavors than is logically possibleSelf-allocations sum to more than 100%Why? Because ones own contributions are more available than those of others
AvailabilityExperimental evidenceMarried couples asked to allocate responsibility for:Positive events: Making breakfast, planning activities, shopping for family, making important decisionsNegative events: Causing arguments, causing messes, irritating spouseResults: Overclaiming occurred for 16 of 20 activitiesEquivalent overclaiming for positive and negative eventsRoss & Sicoly, 1979; Kruger & Gilovich, 1999
AvailabilityWhat is availability? Two possibilities:1. Number amount of information generated2. Ease the ease with which information can be generatedIconic study teased them apart:Participants were asked to evaluate their own assertivenessBy generating either 6 (easy) or 12 (hard) examples of assertiveness or unassertiveness
Availability: number versus easeMoral: Ease influences judgments sometimes in spite of numberSchwarz et al., 1991
Chart1
6.35.1
5.26.2
Six examples
Twelve examples
Higher #s = more assertive
Sheet1
Six examplesTwelve examplesSeries 3
Assertive examples6.35.12
Unassertive examples5.26.22
Category 33.51.83
Category 44.52.85
To resize chart data range, drag lower right corner of range.
RepresentativenessDetermining class inclusion or likelihood by similarity:A member ought to resemble the overall categoryAn effect ought to resemble or be similar to the causeAn outcome ought to resemble the process that produced itLike goes with likeOften easier to assess similarity than probabilityDoes he look like an engineer?Does it look like it could cause a clogged artery?Does it look like a random sequence?
RepresentativenessLeads to several classic judgment errorsConjunction fallacyMisperceiving randomnessRegression fallacy
The Linda problemLinda 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 criminal justice, and also participated in anti-nuclear demonstrations.Rank likelihood that Linda is:A teacher in elementary schoolActive in the feminist movementA member of the League of Women VotersA bank tellerAn insurance salespersonA bank teller and active in the feminist movement
The Linda problemClass data (rankingslower numbers mean more likely):
Active in the feminist movement: A bank teller: Active in feminist movement a bank teller:
Representativeness: Conjunction fallacyJudging the conjunction of two events to be more probable than one of the constituent elementsFeministsBank tellersP(A & B) > P(A) or P(B)/
Conjunction fallacyHow much would you be willing to pay for a new insurance policy that would cover hospitalization for:
1. Any disease or accidentMean = $89.102. Any reasonMean = $41.53 Johnson et al., 1993
Conjunction fallacyHow much would you be willing to pay for flight insurance (1 flight to London) that covers death due to:
1. Any act of terrorismMean = $14.122. Any reasonMean = $12.03Johnson et al., 1993
Representativeness: RandomnessEffects should resemble the process that produced themIf something is random, it should look random
What does random look like?HTHHHTTTTHTHHTTTHHHTHHTHTHTTTHHTHTHTTHHHTH
The hot handIf Im on, I find that confidence just buildsyou feel nobody can stop you. Its important to hit that first one, especially if its a swish. Then you hit another, andyou feel like you can do anything.--Lloyd Free (a.k.a. World B. Free)
The hot handThe belief that success breeds success, and failure breeds failure100 basketball fans91% thought player has a better chance of making a shot after having just made his last two or three shots than he does after having just missed his last two or three shotsGiven a player who makes 50% of his shots, subjects thought that shooting percentage would be61% after having just made a shot42% after having just missed a shot84% thought that its important to pass the ball to someone who has just made several shots in a rowGilovich, Vallone, & Tversky, 1985
The hot handCalculate probability of making a shot after missing previous 1, 2, or 3 shots and after making previous 1, 2, or 3 shotsGilovich, Vallone, & Tversky, 1985
What the hot hand results meanThe independence between successive shots, of course, does not mean that basketball is a game of chance rather than skill, nor should it render the game less exciting to play, watch, or analyze. It merely indicates that the probability of a hit is largely independent of the outcome of previous shots, although it surely depends on other parameters such as skill, distance to the basket, and defensive pressureThe availability of plausible explanations may contribute to the erroneous belief that the probability of a hit is greater following a hit than following a miss. Gilovich et al., 1985, pp.312-313
Regression to the mean
Chart1
1
2
3
1
2
3
4
1
2
3
4
5
2
3
4
5
3
4
5
Y-Value 1
Sheet1
X-ValuesY-Value 1
11
12
13
21
22
23
24
31
32
33
34
35
42
43
44
45
53
54
55
The SI jinx
The SI jinxIn sports (the SI jinx, the sophomore slump, rehiring the interim manager, etc.)In education (the illusory superiority of punishment over reward)In medicine (why its so easy to believe that a worthless remedy really works)In politics (be careful about taking office during an economic boom or a drop in crime)
Part 2: BiasesOverconfidence and its causes
Overconfidence in social predictionsWould the target personPrefer to subscribe to Playboy or the New York Review of Books?Describe his/her lecture notes as neat or messy?Say s/he would pocket or turn in $5 found on the ground?Object when the experimenter referred to him/her by the wrong name?Comb his/her hair before posing for a photograph in the lab?How confident are you in your answer (50-100%)?Mean confidence: 75.7%Mean accuracy: 60.8%When 100% confident, accuracy = 78.5%!Dunning et al., 1990
Overconfidence in self predictionsWill youVisit San Francisco more than 3 times this year?Participate in the dorm play?Drop a course?Question your decision to attend Stanford?Become best friends with your roommate?Visit a friend more than 100 miles away?Get a new boy/girlfriend?Overall confidence: 82.3%Overall accuracy: 68.2%When participants were 100% confident, they were correct only 77.4% of the time!Vallone et al., 1990
Causes of overconfidenceHindsight biasMotivated and non-motivated confirmatory thinkingConfirmation biasWishful thinkingNave realism
Nave realismYou drive up to San Francisco with friends to celebrate the end of the quarter. The plans include dinner and then some entertainment afterward.How much money will you personally spend on the dinner?You receive a telephone call from a survey firm. You initially agree