The Psychology of Judgment and Decision Making

1
The Psychology of Judgment and Decision Making
2

• Part 1
• Heuristics and Biases
• How they affect peoples judgments and decisions
• How they affect researchers
• How they affect MIS

3
Definition Heuristic and Bias
people rely on a limited number of heurisitc
principles which reduce cimplex tasks of
assessing probabilities and prediciting values to
sijmpler judgmental operations. In general, these
heurisitcs are quite useful, but sometimes they
lead to severe and systematic errors. (from
Tversky, Kahneman, Science, 185,
p.1124-1131.) - People are systematically
biased
4
Section 1 Perception, Memory, and Context
Biases
5
A Foolish Consistency is the Hobgoblin of Small
Minds- Ralph Waldo Emerson
6
Previous Beliefs and Expectations Affect
Perception
It Looks Red
7
People Reconstruct MemoriesGuess 4 Numbers
Not Bad Guesses! My Numbers are 1, 2, 5, and 16
I Guessed 2, 3, 6 and 15 I was Right!
8
Cognitive Dissonance
Save The Red Elephants
It WAS Red
Youll Sign My Petition, Wont you?
Joe
I saw red Elephant, X
9
The Halo Effect
That Elephant Has a Big Smile I bet Hes Red
10
Contextual Clues
Small Elephant
Big Elephant
11
Parables of context biases and Purple Red
elephants

• People see what they expect to see (Mr. Red)
• People reconstruct memories (I was right)
• People try to be consistent (It was red)
• Nearby things affect perception (relativity)
• People believe good things about good people (The
  halo effect)

12
Escaping Context Biases

• Take note of context
• Know yourself
• Consider the possibility of alternative outcomes

13
How to get the results you want from an
experiment, capitalizing on Perception, Memory
and Context
(A roll play emphasizing the impact perception,
memory, and context can have in a research
setting)
14
Halo Effect

• Their Tool
• Host uses bad grammar
• Host has unkempt appearance
• Hosts uses poor body language (slouches)

• Our Tool
• Host smiles and conveys excitement about the new
  tool
• Host is nicely dressed
• Host greets people as they enter the experiment

15
Create Expectations

• Their Tool
• Host says there is a chance our tool might work
• Host says we think our tool produces marginally
  better search results

• Our Tool
• Host expresses confidence about the performance
  of the new search tool.
• Host says we are excited about how effective
  our tool is

16
Control Comparatives

• Their Tool
• Begin with a warm-up exercise using
  state-of-the-art technology that is well
  polished.
• Show the group the best of breed searching tools
  available.

• Our Tool
• Run the warm-up exercise on an old, slow computer
  attempting an impossible task.
• Provide a poor tool to achieve the task.

17
Get Them to Commit to Your Result

• Their Tool
• Conduct the following pre-experiment survey
• I am a good searcher who gets good results.
• I am a jerk
• (Please choose only A or B)

• Our Tool
• Conduct the following pre-experiment survey
• I learn quickly and do well with new tools.
• I am a jerk
• (Please choose only A or B)

18
Give Them Cues About How They Did

• Their Tool
• Host frowns when he/she looks at the screen.
• Host shakes head and makes low grunting noises.

• Our Tool
• Host smiles as he/she looks at the participants
  screen.
• Host takes notes by noticeably checking items off
  a list while nodding head.

19
Give Them Time to Reconstruct Memory

• Their Tool
• Host ends the session by saying Wow that took a
  long time.umcome back next week to fill out the
  survey.

• Our Tool
• Host ends the session by saying we were doing so
  well, we lost track of time. Oh well, just come
  back next week to fill out the survey.

20
Section 2 Survey Question Biases
21
Survey Questions Biases
Question phrasing, what answer choices are
presented and how they are presented...
!!?!!
??!??
22
On September 27, 2001, President Bush put forth
the UN Hostile Powers resolution forbidding UN
Humanitarian Economic aid to Afghanistan. Was
President Bush within his powers as the President
of the US to enact such a United Nations
Resolution?
a. President Bush was definitely within his
power. b. President Bush overstepped his
authority.
Will you vote in favor of a special finance
package costing the average taxpayer 475 to
finance Operation Enduring Freedom?
a. Yes, since it is likely Congress with provide
for another tax rebate when the economy turns
around.b. No, 475 in taxes negates the recent
tax rebate.I would need to understand more about
the finance package.
23
pseudo-opinion
On September 27, 2001, President Bush put forth
the UN Hostile Powers resolution forbidding UN
Humanitarian Economic aid to Afghanistan. Was
President Bush within his powers as the President
of the US to enact such a United Nations
Resolution?
Framing
a. President Bush was definitely within his
power. b. President Bush overstepped his
authority.
Ordering
Forbid vs. not allowing
Attitude/Behavior
Will you vote in favor of a special finance
package costing the average taxpayer 475 to
finance Operation Enduring Freedom?
a. Yes, since it is likely Congress with provide
for another tax rebate when the economy turns
around.b. No, 475 in taxes negates the recent
tax rebate.I would need to understand more about
the finance package.
Gain/Loss
Filtering
24
Research Life Cycle Observation

• Through the use of surveys, observations verify
  or contradict MIS research hypotheses.
• If the survey is biased, the research is
  nullified.
• Valid theories could be unnecessarily quashed.
• If the survey is manipulated to reinforce the
  hypothesis, the research will not withstand
  further investigation.
• This will hider the acceptance of MIS as a
  discipline

25
Section 3 Models of Decision Making
26
Expected Utility Theory

• --A normative theory of behavior
• --Rational decision making principles
• Ordering of alternatives
• Dominance
• Cancellation
• Transitivity
• Continuity
• Invariance

Utility
27
The Allais Paradox-- Violation of Cancellation
Principle
Alternative A 1,000,000 for sureAlternative
B A 10 chance of getting 2,500,000, an 89
chance of getting 1,000,000, and a 1 chance
of getting 0
Alternative A An 11 chance of getting
1,000,000, and an 89 chance of getting
0 Alternative B A 10 chance of getting
2,500,000, and a 90 chance of getting 0
The paradox is that anyone choosing Alternative
A in the first situation should also choose
Alternative A in the secondotherwise, the
Cancellation Principle is violated.
28
Intransitivity

• If the difference in intelligence between any
  two applicants is greater than 10 points, choose
  the more intelligent applicant. If the
  difference between applicants is equal to or less
  than 10 points, choose the applicant with more
  experience.

29
Prospect Theory--Improvement

• replaces notion of utility with value
• different value functions exist for gains and
  losses
• (Endowment Effect)

30
Prospect Theory--Improvement

• Peoples tendency to overweight small
  probabilities.
• A reduction of the probability of an outcome by a
  constant factor has more impact when the outcome
  was initially certain than when it was merely
  probableCertainty Effect

31
Other Alternatives to Expected Utility Theories

• Satisficing
• Regret Theory
• Multi-attribute Choice
• Non-compensatory Strategies

32
Conclusions in MIS Field

• Take caution in making assumptions and
  principlesthey are easily violated.
• Design as many experiments as you can, and
  consider as many conditions as you can to make
  your principle complete.

33
Section 4 Representativeness Availability Risk
Anchoring Randomness Causation Attribution
34
Heuristics And Biases

• People rely on heuristic when facing complicate
  decision
• Advantage normally optimal answer reduce
  time and energy
• Disadvantage lead to biases and inconsistency
• Focus the process reaching the
  conclusionconsequent biases

35
Representativeness Heuristic
Definition Judge probability by the degree to
which A is representation of B

• Example1
• P(A) P(AB)
• Dont be misled by highly detailed scenarios

• Example2 The law of small numbers
• gamblers fallacy vs. the hot hand
• Remember that chance is not self correcting
• Dont expect too much alternations

36
Representativeness Heuristic (2)

• Example3
• use base rate information only when its
  consistent with their intuitive theories of cause
  and effect
• Whenever possible, pay attention to base rates

• Example4
• tendency to make nonregressive predictions
• Dont misinterpret regression toward the mean

37
Availability Heuristic
Definition judge probability by the ease with
which the instance can be brought to mind
Examples media, difficult to generate, visualize

• The limits of imagination
• difficult-to-imagine
• extremely negative outcome

38
Vividness Heuristic

• Definition
• concrete or imaginable emotionally interesting
  or exciting close in space or time
• Examples individual testimonials
• it does exist, but is limited(time, laboratory
  research)
• how to conquer explicitly comparing

39
Probability and Risk

• Bias in heuristic
• More statistics Bayes theory
• Wishful thinking happens to me? Never!
• Compound events Welcome to the real world!
• Each of all 500 components has the rate of 99
  working great, .now you feel better?
• I am the expert..so what?
• Perception of risk are strongly biased in the
  direction of preexisting view.

40
Probability and Risk (2)

• Managing the information system
• Deal with new information
• Maintain accurate records
• Avoid conservatism
• Avoid convenience
• Be objective
• Be aware of wishful thinking
• Break component events into simple events
• Realize different perspectives
• Reduce risk in system building

41
Anchoring and Adjustment

• Introductory example Average January temperature
  in Pocatello, Idaho
• The insufficient adjustment up or down from an
  original starting value, or anchor.
• Very Robust
• Works with extreme anchors
• Examples
• Real estate
• Software development
• Give me a preliminary estimate

42
Anchoring and Adjustment in Research

• Possible problems
• Anchoring on previous research
• Anchoring on a hunch
• Solutions
• Be skeptical of previous research
• Be thorough with methodology

43
Perception of Randomness

• Coincidence
• Odds of particular episode happening are low, but
  odds of some similar episode happening somewhere
  are high
• Examples
• Dr. Bookers story of meeting Dr. Nunamaker
• Which one is more random?
• 0111001000111001011011
• 0010101010110101000101

44
Randomness and Research

• Finding patterns in research data that dont
  exist
• Stick to the methodologies and statistical
  analysis
• Coincidences happen
• Take advantage of those coincidences that happen
• Increase the odds of coincidences by networking

45
Correlation, Causation and Control (CCC)

• Correlation whether two variables are related.
• Illusory correlations
• Common sense dominated
• bacon-tiger vs. bacon-egg
• Invisible correlations
• Absence of expectation
• meat consumption vs. colon cancer

46
CCC (2)

• Causation
• Correlation and Causation
• Control
• Self confident I can control myself better!
• Helps that harms connections between health and
  a sense of control

47
CCC (3)

• Take-aways
• Bias in information colleting and decision making
• Focus on more than confirming and positive cases
• Observation vs. expectation
• Distinguish between correlation and causation

48
Attribution Theory

• Definition Attribution theory is a psychological
  theory about how people make causal
  attribution, or explanations for the causes of
  actions and outcomes

49
Important Exception

• Attribution theory works most of the time, but
  there are several important exceptions
• When people disregard consensus information
• Salience information has more impact
• Behavior is the most salient thing in a social
  setting

50
Important Exception (2)

• Difference in focus between actors and observers
• People rely heavily on the most salient factors
  at the time
• Other attributional biases

51
Methods to Avoid Attributional Pitfalls

• Pay close attention to consensus information
• Ask how you would have behaved if thrust into the
  same circumstance

52
Section 5 Biases in Decision Making in
Social and Group Contexts
53
Group Influences

• Social facilitation (Mastery vs. Performance,
  1965)
• Social loafing (Work in group vs. Work alone,
  1965)
• Bystander intervention (Are you above 9?)

54
Group influences (2)

• Taking cues from those who are similar
• Social analgesia (Influenced by perceptions of
  others)
• GroupThink loyalty and pressures to conform
• Minority influence on majority.

55
Biases in Group Research

• Individual-level biases exist in groups as well
• Preexisting tendencies are amplified
• The less permissive the leader is, the worse the
  group performs
• The group performs no better than the best
  member
• Brainstorming in group is less effective than
  conducting with people individually.

56
Section 6 Common Traps
57
Common Problems

• Overconfidence
• Self-fulfilling prophecies
• Behavioral Trap

58
Overconfidence (OC)

• A belief in ability or understanding that
  adversely affects a decision
• Space Shuttle Challenger
• Pearl Harbor
• Heuristics
• Measure or assessment of confidence level
• Accuracy of decision
• Bias
• Discrepancies between accuracy and confidence

59
Overconfidence Causes Impacts

• Possible Causes
• When judgment difficult (no right answer)
• Ignorance (initial phases of research)
• Impacts
• Extreme
• Overconfidence when essentially certain
• Not just a consequence of underestimating risk or
  misunderstanding
• Social
• Wrongful convictions
• Societal/moral losses due to decision maker
  overconfidence
• more confidence belief in what U are doing

60
Overconfidence Fixes

• Calibration
• Bringing decision makers confidence in line with
  accuracy
• People can learn
• Lots of feedback after decision making
• Takeaway
• Stop to consider reasons why your judgment
  might be wrong

61
Overconfidence and MIS

• Design system - feel good - overlook bug
• Literature review - feel informed - faulty
  conclusion

62
Overconfidence and MIS

• May reduce flexibility in judgment
• May Reduce
• Dissertation - ask questions to learn - ask
  yourself questions to verify your thoughts

63
Self-Fulfilling Prophecies (SFP)

• Confirmation Bias
• A positive testing pattern (confirmation bias)
  leading to difficulties in decision making

Total Decision space
Answer
Biased decision space
64
SFP Model
Source http//www.mhhe.com/socscience/social/iban
k/set-1.htm
65
Pygmalion Affect (SFP)

• A false definition of a situation causing a new
  behavior making an original false conception true
• Cause
• Stereotypes, perceptions
• Advantage (if used properly)
• Treat people what they ought to be and you help
  them become what they are capable of being
• Disadvantage
• Self fulfilling prophecies, if not careful

66
SFP Impacts

• Pygmalion Affect
• Favoritism in classroom by teacher towards
  gifted students
• Gifted students excel
• Perceptions of women by men
• Male subject presented with snapshot of
  attractive women
• and unattractive woman
• Attractive women favored after phone conversation

67
SFP Impacts (2)

• Interview process based on race
• Minorities performing differently during
  interview process when interviewer is postured
  defensively or with distance

68
SFP Fixes

• Focus on motivational factors
• Warning interviewers about questions that
  indicate closed-mindedness
• Can be used to motivate problem students
• Request disconfirming information during
  interviews
• Similar to proving null-hypothesis before
  conclusion??
• Possibly by asking asking why judgments might be
  wrong

69
SFP and MIS

• Research question produces results found in
  context of original research question
• Avoid testing techniques/methodologies that are
  too narrow confirmation bias
• System design
• Research hypothesis
• Favoritism based on expected or forecasted
  performance Pygmalion affect
• Design technique
• Upcoming technology

70
Behavioral Traps

• A course of action that starts out as promising,
  but later becomes undesirable
• Traps
• Avoiding potentially beneficial behavior
• Waiting on hold for tech-support
• Exercise
• Counter-Traps
• Potentially harmful behavior
• Avoiding chores
• Cigarette smoking

71
Behavioral Trap Categories

• Time delay
• Instant gratification vs. long term
  consequences
• Smoking one more cigarette
• Ignorance
• Result of action not understood or predictable at
  start of action
• Smokers in the 19th century
• Starting a new job only to find out that it sucks.

72
Behavioral Trap Categories (2)

• Investment
• Prior investment of resources (e.g., time, money)
  leading to choices not normally made
• Vietnam war
• R D investment (time, money) only to be out
  done by competition
• Should project continue????
• Lead to sunk-costs

73
Behavioral Trap Categories (3)

• Deterioration
• Similar to Investment traps, but benefits and
  costs change over time

74
Behavioral Trap Categories (4)

• Collective
• Quest for self interests having negative impacts
  on group
• Two criminals offered choice confess or go to
  jail
• 1 Confesses no jail - partner gets 10 years
• 2 Confess - both get 5 years
• No confession - both get 1 year

75
Behavioral Trap Fixes

• These traps often dont last long
• People learn and move on
• Corporate leaders stop wasting money
• Girlfriend dumps boyfriend (he calls old
  girlfriend)
• Set limits in advance
• Use limits to assess continuation

76
Behavioral Traps and MIS

• Starting system design before jumping to code
• Gather facts before making conclusions
• Identifying scope of project before starting and
  assessing status and progress regularly
• Identifying that research question may not take
  expected path and adapting.

77
Part 2 Decision Making and Heuristics in a
Complex World
78
Heuristics in a Complex World

• Humans are boundedly rational given the amount
  of information necessary for rationality and
  human cognitive abilities
• Future is difficult to predict, both externally
  (states of the world) and internally (future
  preferences)
• - Humans use simple heuristics to satisfice
  rather than optimize

79
Simple Heuristics that Make Us Smart
Heuristics are frugal get away with little
information fast dont need long computation
cheap dont need costly information search,
storage, and computation robust perform better
than optimal regression on new data sets less
affected by noise
80
Overconfidence Revisited

• Questions in overconfidence studies are not
  randomly sampled from the environment, but
  selected by researchers
• Frequency judgments of correct answers are quite
  well-calibrated
• - Should researchers focus on finding biases or
  explaining normal functioning?
• - Which perspective is better for MIS
  researchers?

81
Heuristics and Biases A Paradigm

• Tversky and Kahnemans heuristics and biases has
  been the dominating paradigm in JDM research for
  about the last 30 years
• What are the benefits and costs of a strong
  paradigm?
• When does a discipline need a paradigm shift?
  (When researchers get bored?)
• What is the ruling paradigm in MIS? How strong
  is it? What is its effect?

82
New Perspective in JDM Emotions Neuroscience

• Risk as Feeling
• Regret Avoidance
• Mood and Decision Making
• Descartes Paradox
• Emotion as Information

83
(No Transcript)