Decision-Making

Question 1

judgement and decision-making

Answer 1

Combines cognitive psychology, social psychology, and behavioural economics

Question 2

what is a decision? what is a risky decision? what are parameters of decisions?

Answer 2

• Anything with >1 response option
• Risky decisions: those with emotional outcomes (good things and bad things)
• Decisions can vary along a number of parameters:
• • Size of the outcomes (gain and loss)
• • Probabilities of the outcomes occurring (judgment)
• • Delay to receiving outcome
• • Effort to obtain outcome

Question 3

what is risk?

Answer 3

• hard to define
• 3 ways:
• • risk as variance (economics)
• • risk as hazard (psych/med)
• • risk vs. uncertainty (coined by Knight)

Question 4

risk as variance

Answer 4

• economic viewpoint
• Ex. If you flip a coin and heads gains you $10 and tails gains you $90, or you flip one where heads gains you $40 and tails gains you $60, the first coin is more risky because there’s a bigger discrepancy between 10 and 90, even though you’ll gain either way

Question 5

risk as hazard

Answer 5

• potential for negative consequence (psychology and medicine)
• Ex. If you flip a coin and heads loses you $10 and tails gains you $50, or you flip one where heads gains you $10 and tails gains you $30, the first coin is more risky because there’s a risk of losing money

Question 6

risk vs. uncertainty

Answer 6

• term coined by Knight
• Decisions under risk = known probability distribution (ie. You know your chance of winning the lottery is 1 in 14 million, for example)
• Decisions under uncertainty/ambiguity = unknown probability distribution; must be estimated or learned through experience (ie. You have no idea what your probability of winning is if you’re sports betting)

Question 7

expected value

Answer 7

• EV = [probability x value] for all outcomes for an option
• Choose option that maximises expected value
• Example: Will you bet $5 for chance of winning $50 by rolling snake eyes?
• • Statistical chance of snake eyes is 1/36 (1 in 6 chance on one and 1 in 6 chance on the other)
• • EVa: (1/36 x 45 -> what you’d profit) + (35/36 x –5 -> what you’d lose) = -3.6
• • EVb: 0 (you won’t win or lose anything if you walk away)
• • EVb is better option -> walk away

Question 8

expected utility

Answer 8

• Life’s not all about money -> if an apple and an orange cost the same amount of money, but you like apples better, apples have higher utility
• A common currency
• Example: should I buy a lottery ticket?
• • Options:
• – Buy: Win jackpot -> Many “utils”; Lose -> Negative “utils”; Feeling hopeful that you might win -> some “utils”
• – Don’t buy: Anticipatory regret (wondering if the numbers you always play are going to win) -> negative “utils”

Question 9

The Bernoulli Effect

Answer 9

• “gain of 1000 ducats is more significant to the pauper than the rich man”
• Diminishing marginal utility with increasing value

Question 10

Violations of expected utility

Answer 10

• ex. Tropical Flu problem: choose to save 200 people or take a 1/3 chance of saving 600 (2/3 of saving nobody); or choose not to save 400 people or take a 1/3 chance of saving everyone (2/3 chance 600 people die)
• The “Framing” effect: All four choices are mathematically identical in their EV -> according to economists, people shouldn’t have a strong preference
• • However, psychologists found that in the gains version, most subjects are risk averse (A>B) and in the risk version, most subjects are risk preferred (D>C)

Question 11

How to we explain the framing effect in the tropical flu problem?

Answer 11

• the value function
• In the gains domain, the curve is concave, in the losses domain, the curve is convex; curve is steeper in the loss domain (-> loss aversion)

Question 12

prospect theory: 3 ingredients

Answer 12

• The Value Function (relating objective value to subjective value)
• Decisions made relative to a “reference point” (the origin)
• Probability converted to subjective probability via the Weighting Function (an inverted S shape)
• • Meaning that people tend to overestimate unlikely things and underestimate likely things -> ex. We over-estimate rare causes of death and underestimate common causes
• • People are very accurate at judging probability when it’s 0, 1, or around 0.4-0.5 (when they cross over)

Question 13

Heuristics

Answer 13

• Although economic approaches to decision-making argue that people maximize expected value (utility) by combining gains and losses with subjective probabilities to make sound decisions
• However, much of the time people use heuristics: short-cuts that avoid algorithmic processing
• Heuristics are fast, usually give an answers that is good enough and well-suited to the everyday world, but can sometimes generate silly responses

Question 14

Bayes’ Rule

Answer 14

• Gives us a way to revise our probability estimates in light of new information
• Takes into account the Initial estimate/prior probability/base rate, the hit rate, the false alarm rate

Question 15

Bayes’ rule: breast cancer example

Answer 15

• People tend to overestimate chances of the woman having cancer, but if you actually crunch the numbers using Bayes’ rule, it’s much lower
• Why are estimates so high?
• • People show base rate neglect – they ignore the low initial probability
• • Instead, they make their estimate using the information (the test accuracy) that appears representative or what they are trying to do
• • The representativeness heuristic: probability of an event is estimated from a similar situation with a known probability
• Often better to use frequencies to crunch numbers rather than probabilities

Question 16

The Gambler’s Fallacy

Answer 16

• Imagine you are betting on a colour in roulette -> last 4 outcomes are red, red, red, red
• What do you bet on? Many will say black because they feel like it’s “due” -> this is the gambler’s fallacy (failing to understand that each spin is independent of the last -> the roulette wheel isn’t taking all of those reds into account)

Question 17

The Gambler’s Fallacy as the representativeness heuristic

Answer 17

• People expect short sequences of outcomes to be representative of the overall population; runs of one outcome appear non-random
• In reality, “chance is lumpy”

Question 18

Heuristics in blackjack (Wagenaar study)

Answer 18

• Wagenaar field study of backjack players in the casino: How do players respond to runs of wins and runs of losses?
• Most increased their bet after runs of losses, and decreased their bet after runs of wins -> representativeness
• Some showed the opposite, increasing bets after a win and decreasing bets after losses -> availability heuristic

Question 19

Illusory correlations

Answer 19

• Seeing a correlation (or relationship) where none exists
• • Ex. Chapman study: random pairs show up on screen equally often, but participants rank the 2 “familiar” pairs (Bacon and eggs; lion and tiger) as showing up more often -> availability heuristic
• • Ex. You remember the instances where you wore your lucky t-shirt and won much more easily than when you wore the lucky t-shirt and lost, or won without wearing the t-shirt (Related to availability heuristic and confirmation bias)

Question 20

Hindsight bias

Answer 20

• The “you knew it all along” effect – an event is more predictable/was more likely after it has occurred
• • Ex. Fischoff study: 4 groups estimated likelihood of 4 different outcomes of a British conflict in Nepal, estimates were higher after they found out result (memory of their knowledge shifts their memory of the estimates)

Question 21

The bias blind spot

Answer 21

• We believe we’re not influenced by bias, but that others are
• This happens in gambling -> odds apply to the other gamblers, but not to me

Question 22

Overcoming biases/”debiasing”

Answer 22

• Education on gambling and statistics
• Experiential debiases (ie. See how many throws it takes you to roll “snake eyes” in an intervention for explaining rare events