Lecture 8

Question 1

What are the two sources of DM mistakes?

Answer 1

• Assessing subjective expected utilities
• Assessing subjective expected probabilities

Question 2

How to measure utility?

Answer 2

• Absolute vs relative: hard vs easy
• Easier to answer relative questions e.g which would you prefer rather than rate this on a scale
• Good to place things in a rank, even random ones to give you a more precise idea
• On a scale from -100 to +100
• Anchors matter e.g -100 = hell, +100 = heaven
• E.g utility of a burger on the scale of heaven/hell does not make sense - too extreme = compresses judgements into a smaller range - hard to differentiate, use more of the scale when you make it more precise as it represents more

Question 3

How to measure utility: framing?

Answer 3

• Prospect theory: people treat gains and losses differently
• Gains: risk averse, losses: risk seeking
• People treat gains and losses relative to a set point in relation to an absolute measure
• Framing tends to be non-normative: if utility measurements are differently questioned, do we get a different answer

Question 4

How to use preferences to measure utility?

Answer 4

• Utility of best option is 100 (arbitrary)
• Utility of worst thing is 0
• Utility of middling outcome is 100*p
• Imagine you are a lecturer for certain versus p chance of king of the world and 1-p being broke and homeless e.g lecturer for certain but chance of king = 1% and broke = 99%, what to choose until you reach the tipping point

Question 5

What are substitution measures of utility?

Answer 5

• How much money would it take to give up X

Question 6

What was the pain thought experiment with money?

Answer 6

• Memory for pain is dominated by peak pain and final pain
• Memory for past utility can be badly distorted
• Kahneman Pain Studies

Question 7

What is probability?

Answer 7

• From 0 - 1
• Two theoretical frameworks: probabilities are statements about event frequency e.g dice rolling
• Probability judgements are statements about degree of belief in the occurrence of an event e.g beliefs exist where frequencies are unclear or ambiguous = p(you will be married by 27)

Question 8

What is coherence?

Answer 8

• Not violating the rules of probability
• State probabilities must sum to 1
• Subsets cannot be more probable than the sets they are in
• States must be mutually exclusive and exhaustive

Question 9

What is Correspondence?

Answer 9

• Accurately matching subjective to objective probability
• Risk overconfidence effects: tendency to estimate probabilities closer to 0 or 1 than they are
• Risk typicality effects: tendency to over-estimate the probability of typical events

Question 10

How can frequencies clarify degree of belief?

Answer 10

• Direct Assessment: p(Labour next PM)
• Frequency Assessment: how many PM changes have been leader of opposition
• Frequency effect: People tend to be better at reasoning with frequencies than probabilities
• Reduced conjunction fallacy and reduced base-rate neglect

Question 11

What is the rare event fallacy?

Answer 11

• People tend to overestimate the probabilities of rare events e.g rain in cardiff vs die because of a dog
• Anchoring unknown events relative to rare events with know odds can be useful
• Availability heuristic can inflate the probability of rare events

Question 12

How to avoid lazy probabilities?

Answer 12

• e.g probability of alien humans being bipedal is not 0.5
• Decomposing state probabilities can be useful
• Careful of subadditivity errors