Lecture 10

Question 1

Knight’s three probabilities

Answer 1

(1) a priori probability
(2) statistical probability
(3) estimated probability

Question 2

a priori probability

Answer 2

An absolute, mathematically calculated probability

Question 3

Statistical probability

Answer 3

Empirical evaluation of the probability from past instances

Question 4

Estimated probability

Answer 4

Estimation when there are too few instances (rare) to properly determine statistical probability from.

Question 5

Knight’s theory of risk

Answer 5

There are two different kinds of uncertainty/risk: measurable and unmeasurable. These produce different behavior

Question 6

Risk vs. Uncertainty vs. Ambiguity

Answer 6

Uncertainty: state where future outcome of decision is unknown. Incapsulates both risk and ambiguity

Risk is measurable uncertainty, and has a probability that is calculable or quantifiable

Ambiguity: state where it is not possible to measure or say the likelihood of each outcome; unquantifiable probability

Question 7

Certainty effect

Answer 7

People like certainty; as soon as you remove it, things like the Allais Paradox go away and people choose the higher EV option. People like BOGO more than 50% off two although it is the same.

Question 8

Pseudocertainty

Answer 8

People prefer a vaccine that has 100% of preventing 85% of HPV strains over a vaccine that is 85% effective, even though same thing. We like 100%.

Question 9

Common-ratio effect

Answer 9

Preference reversals when probabilities of two options of a pair are each divided by a common number. So 100% to 25%. This is driven by over/underweighting of probability (prospect theory) and by the certainty effect

Question 10

Ambiguity aversion

Answer 10

We prefer known uncertainty over unknown uncertainty - risk better than ambiguity. Ellsberg’s paradox - people would rather bet on known odds than ambiguous odds of equal size. However, if we perceive confidence we tend to prefer ambiguous odds

Question 11

Decisions from description vs. experience

Answer 11

From description, we tend to over-weight low probabilities (like dying from a vaccination). From experience, we tend to under-weight low probabilities

Question 12

Description-experience gap

Answer 12

Shifts in choice depending on if we learn about probabilities from description or experience

Question 13

Cause of description-experience gap

Answer 13

Heuristics probably lead to underweighting rare events. Could be maximax, maximax, minimax, lexicographic, or natural mean heuristic

Question 14

Interaction of description and experience

Answer 14

People who are given descriptions but also allowed to sample thru experience look like pure experience learners - experience may override