Values, Choices and Frames
Session 10Decision Making and Risk Sp2006
Role of Memory and Processing Biases in Decision Making
In-Class Exercise – Generating Probability Estimates
Generate examples of probability estimation situations in which the estimates could be biased (previous lecture), and come up with ways to correct such estimates.
Output: A table with three columns, first column, the name of
the bias, second column, a business situation which exemplifies the bias (this has to be somewhat descriptive), third column that makes specific recommendations about how the bias can be avoided.
New Product Development Situation The product development team for NexTech Inc., a toy
manufacturer and marketer is in a new products ideation meeting featuring several participants.
The goal is the following: Identify as many new product ideas that will go into the
‘decision option’ pool. They have to come up with decision criteria, and
importance assessed to each criterion thereof, that will help them evaluate the decision options they generate.
How should they go about the task? Specifically, what would limit/expand the number of new
ideas generated? What would affect their ability to go closer to/away from
the true decision criteria?
Part-List Cueing Effect Time 1:
Learn a list of items in a category
Time 2: Unrelated activity
Time 3: Split into two groups.
One group studies a random selection of half the items of again (Re-Study Group)
The other group does not (No Study Group). Time 4:
Unrelated activity
Time 5: Both groups asked to recall as many items from the original list.
Results:
For items studied at time 3 Re-Study Group > No Study Group
For items not studied at time 3Re-Study Group < No Study Group
Part-List Cueing
If there are “n” options in an option space, S.
Let rn be the number of options recalled from memory; rn ≤ n.
If a few options, say, c, are provided as a memory cue, then, rc, the number of items recalled from S when ‘c’ items are provided, will be such that: rc << rn - c.
How Structure Affects Importance of Decision Criteria
ProblemsPercentage Choosing Each AnswerOpen Question Closed Question
Quality of Public Schools
Legalized Abortion
Energy Shortage All Other
Responses
1100
98
321486
40Source: Schumann and Scott (1987)
What is Important and What is not? Highly affected by availability.
Factors that are listed, get inflated importance scores. Importance of factors generated by pulling
from memory tends to correlate better with true importance.
Take away? For idea generation For importance assessment
Elementary Information Processing Model
InputCognitive and
Emotional Processes
Output
What you remember from a list? Time 1:
Learn a list of items in a category
Time 2: Unrelated activity
Time 3: Recall as many items from the
original list.
What factors will Affect What is Recalled? Vivid items Items that came early on in the list. Items that came toward the end of the
list. Self-relevant items. Mood-congruent recall.
Memory Biases in Information Processing
InputCognitive and
Emotional Processes
Output
VividnessSelf-relevancePrimacy-Recency
Mood Congruent Recall
Simulating Competitor Behavior Company A is trying to assess risk
factors for success of a product. It is also concerned about the moves of
Company B. How would their risk estimates vary? How would their ability to process risk
increasing/decreasing factors vary?
Some Issues when Thinking of Others’ Risk Self-Other discrepancy in risk estimates
Vague competitor, i.e., ‘the competition’ tend to be judged more unfavorably than one’s own firm behavior, success probability etc. Individuate ‘the competition’ to a ‘particular
competitor’ Same thing applies to ‘the customer’. It often arises from:
inability to visualize risk increasing factors being realized for one’s own firm, and risk decreasing factors being realized for competing firm.
Selective memory for risk decreasing factors for one’s firm and risk increasing factors for competitor.
Dizziness and Brain Tumor
Which cells of the table are needed in order to determine whether dizziness is associated with brain tumors in this sample of people? (Check all that apply.)
Upper Left Lower Left Upper Right Lower Right
According to the data in the table, is dizziness associated with brain tumors? Yes No Not sure
Brain Tumor
Present Absent
Present 160 40 Dizziness
Absent 40 10
What Information is Needed for Assessing Correlation?
50
55
60
65
70
75
80
85
90
95
D-Present, BT-Present
D-Absent BT-Present
D-Present BT-Absent
D-Absent, BT-Present
Percent indicating quadrant for which information is needed
Illusory Correlation?
0
10
20
30
40
50
60
Related Unrelated
Percent indicating absence/presence of a relationship between Brain Tumor and Dizziness
SuperTire Discounters
Tire A: 40,000 mile warranty, $60.
Tire B: 50,000 mile warranty, $75.
Without changing the warranty and price (or any other attribute) of Tires A and B, how can one affect market share of Brand A?
Decision before Decision One group gets to choose
between: Tire A’: 55,000 and $85 Tire B’: 75,000 and $91
Another group gets to choose between:
Tire A’’: 30,000 and $25 Tire B’’: 35,000 and $49
Now both groups choose between:
Tire A: 40,000 and $60 Tire B: 50,000 and $75
0
10
20
30
40
50
60
70
80
90
Big Bang Small Bang
Lo-LoHi-HiLo-LoHi-HiLo-LoHi-Hi
Background Tradeoff Contrast Effect Pre-decision choice:
Big versus small bang for the buck Very high increase in value for unit increase in money
compared to the
Target choice set: Tradeoff functions will be contrasted.
With big bang for buck as the background, the increase in quality for money in the target set looks anemic by contrast, so choice is for the low quality low price brand.
With small bang for buck in the background, the increase in quality for money in the target set looks very good by contrast, so choice is for the high quality high price brand.
Local Contrast Effect
Suppose people are trying to choose between: Tire A: 40,000 and $60 Tire B: 50,000 and $75
Introduce Tire A’ which is inferior to Tire A, say, 35,000 and $60.
Tire A’ is dominated by Tire A, but not by Tire B.
A significantly greater proportion choose Tire A when the choice is A, A’ and B, than when the choice is A and B.
Review – Session 10 (April 4) Structuring the Decision
Memory Biases Part List Cuing – How available stimuli limit what is retrieved from memory.
Unstructured – private Unstructured – public Structured – public/private
Inflation of Importance Context of discovery/confirmation
Primacy and Recency Vividness
Self-relevance Mood-congruent recall
Processing Biases Self-other discrepancy in risk estimates, ability to visualize risk increasing
and decreasing factors, ability to use them Illusory correlation.
Value and Choices Value Maximization
Independence of Irrelevant Alternatives Tradeoff Contrasts
Background Contrast Local Contrast