Judgment and Decision Making

The Problem

Jack is a 45-year-old man. He is married and has four children. He is generally conservative, careful, and ambitious. He shows no interest in politics and social issues and spends most of his time on hobbies which include home carpentry, sailing and mathematical puzzles.

In a sample of 70 lawyers and 30 engineers, how likely is it that Jack is an engineer?

The Problem

Judgment Processes are notoriously faulty

Decisions are usually based on partial information The solutions to decisions are often ambiguous.

Leads to the use of: Heuristics

Rules of thumb Biases

Stereotypic decisions

What is a decision?

Person must have a goal There must be many ways to satisfy the goal There is a set of options

Consideration set: Set of options being evaluated

Options are evaluated in some way Eventually one of the options is selected

Economic approaches influenced much of the psychology of choice Theories assume people are rational and want to make

the optimal choice in a given setting Rational decision making

What is the optimal choice i.e. best reflects the person’s preferences?

Decisions should be consistent Law of contradiction

Reasoning processes that use the same information should reach the same conclusions

Those that do not are irrational Example: Transitivity

If you prefer A to B, and B to C… Then you should prefer A to C.

Rational models

Expected Value Theory People calculate the potential value of each option Pick the option with the highest expected value

Raffle with 10% chance to win $5.00 EV = .10 * $5.00 = $0.50 Example: Which gamble would you rather play?

A: 20% chance of winning $5.00 B: 30% chance of winning $4.50 EV(A) = .20 * $5.00 = $1.00 EV(B) = .30 * $4.50 = $1.35 Expected value suggests you should choose B

Expected Value Theory

Problem: Not every dollar has the same subjective value

Graduate student: $100 would allow student to eat better food or to buy new clothes

Lawyer: $100 would not need to be spent on necessities

Example: Lotteries People often play the lottery Pay $1.00 for a 1/52,000,000 chance to win

$10,000,000 Expected value of this gamble is less than $1.00

(~$.19)

Expected Utility Theory

Value of an outcome is based on the individual’s goals What can an option be used for?

That is the expected utility of an option

The Expected Utility model: EU = probability of outcome*utilityi

Expected Utility is a rational model Obeys the law of contradiction All choices are transitive

Everything is evaluated relative to a global scale

Expected Utility Theory

Lottery The expected utility of $1.00 may be low

There is not much you can do with $1.00 The expected utility of the prize may be high

You could do a lot with that kind of money The low probability of winning does not

completely outweigh the high utility of the prize There is also even the pleasure in dreaming

about winning

Expected Utility Theory Problem The Allais Paradox Situation 1

A: A 100% chance to win $1,000 B: An 89% chance to win $1,000 A 10% chance to win $5,000 A 1% chance to win $0

Many prefer A Situation 2

C: An 11% chance to win $1,000 An 89% chance to win $0

D: A 10% chance to win $5,000 A 90% chance to win $0 Many of those who preferred A now prefer D, which is inconsistent as the

difference between the two options is the same 1% difference in probability of outcomes in both situations so u(a) - u(b) = u($1k) - .89u($1k) - .10u($5k) - .01u($0) = .11u($1k) - .10u($5k) - .01u($0) and u(c) - u(d) = .11u($1k) + .89u($0) - .10u($5k) - .90u($0) = .11u($1k) - .10u($5k) - .01u($0) so you should either choose a and c, or b and d.

The irrationality of choice

The Allais paradox represents a certainty bias People prefer to avoid winning nothing and will forgo

the likelihood of a larger amount for certain gain Imagine that the US is preparing for the outbreak of

an unusual disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the program are as follows: Program A: 200 people will be saved. Program B: A 1/3 chance 600 people will be saved,

and a 2/3 chance that no people will be saved. People tend to pick Program A

Gains and losses

The previous example suggests people are risk averse for gains They do not want to risk losing a certain gain. What happens for losses?

Imagine that the US is preparing for the outbreak of an unusual disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the program are as follows: Program A: 400 people will die Program B: A 1/3 chance no people will die, and a 2/3

chance that 600 people will die. People tend to pick Program B

People are risk seeking for losses

Influence of Context on JDM

Framing Effects The way we phrase the question matters Contributes to other heuristics and biases People bias toward absolutes rather than probabilities

Would you volunteer? A disease will inflict 20% of the population. A vaccine is

available that protects half of the people that take it.

Two strains of the same disease will each inflict 10% of the population. A vaccine is available that protects everyone against one strain but not the other

Framing effects

Kahneman and Tversky People treat gains and losses differently

Losses loom larger than gains The same situation feels worse when framed in terms

of losses than when framed in terms of gains May not be true in all cultures

Practical application When making a decision, try to frame the options both

in terms of losses and gains. See whether your opinions about the options changes

Context effects

Explanation-based Decision Making Trying to fit evaluation into the domain of explanation Determines the mental models used

Expected utility predicts that each option is evaluated independently of other options

Adding more members to the consideration set should not influence people’s preferences. The attraction effect The compromise effect

Attraction effect

Choice A and B are better than the other along a particular dimension (e.g. price and quality)

Utility theories suggest that the choice of A or B should be unaffected by the presence of a third alternative Their utility does not change

However, the presence of a completely dominated choice (A vs. C, B vs. D) attracts people to the dominating alternative

Compromise effect

Given choice between D and E

Add F Is tops along

dimension 1 Results in more choice

of D

Preference reversals Slovic & Lichtenstein Different measures of preference may lead to different

outcomes A: 11/12 chance to win 12 chips 1/12 chance to lose 24 chips B: 2/12 chance to win 79 chips 10/12 chance to lose 5 chips Some people asked to choose a bet and then ask how much

they would sell the bet for If choose A should sell for more (expected utility must

be higher) Often gave a higher price for B

There seems to be a compatibility effect Making a choice increases the weight given to

probability Giving a price increases the weight given to the money

prize

What else are people doing to make a (bad) choice? Heuristics and biases Cognitive heuristics are natural ways of

thinking, rules of thumb for decision However, they represent oversimplifications

and may lead to bias

Heuristics and Biases Jack is a 45-year-old man. He is married and has four

children. He is generally conservative, careful, and ambitious. He shows no interest in politics and social issues and spends most of his time on hobbies which include home carpentry, sailing and mathematical puzzles.

In a sample of 30 lawyers and 70 engineers, how likely is it that Jack is an engineer (percent)?

Representativeness Judgments based on the degree to which salient features of

an event match those of a parent population Comparing to the “typical”

Conjunction Fallacy

A health survey was taken of a representative sample of all ages. Mr. F. was in the sample. Which is more probable? Mr. F. has had one or more heart attacks

Mr. F. has had one or more heart attacks and is over 55-years-old.

Conjunction fallacy Mistake in believing the conjunction of traits is

more likely than the individual traits

Representativeness

Insensitivity to prior probabilities Judgments based on perceived frequency of

occurrence. Baseline rates cannot be ignored (Bayesian approach)

Insensitivity to sample size With larger samples come more typical situations

Misconceptions of chance Looking random Gamblers’ Fallacy

Thinking that prior outcomes influence the long run of events

Representativeness

Insensitivity to predictability Predictions based on limited info

Illusion of validity Varying degrees of overlap among sources of

information Tendency to treat dependent sources as independent More sources of (dependent) information increases

confidence without increasing predictive accuracy. Misconceptions of regression

Particular outcome, however extreme, may not necessarily mark a significant turn of events

Regression to the mean

How the Cues are Utilized

Availability Heuristic Judgments based on the ease to which

instances come to mind.

__ __ __ __ N __ __ __ __ I N G

Generate words? Frequency?

Availability Heuristic

Bias due to retrievability of instances Easier-to-retrieve info perceived as more

numerous Solo members/Von Restorff effect

Bias due to (in)effectiveness of search set E.g. more words that start with r or have r as

third letter

Availability Heuristic

Illusory Correlation Not correlated or correlation only due to relationship to

third variable Correlated to a lesser extent Correlated in the opposite direction If one event more frequent, something assumed to be

correlated with will be judged accordingly

Counterfactual thinking Availability of alternate explanations

Other biases

Confirmation Bias Tendency to seek or recall information that confirms a

hypothesis (or diagnosis) rather than information that refutes the hypothesis.

Hmmm… I wonder if that’s seen in the sciences at all? May determine what cues are used in judgment Often the source of prejudice

Hindsight bias I knew it all along

Overconfidence Use of norms

What may be the usual case may not apply Ease of imagining possible alternatives

Influence of Context on JDM

Order of Cues Anchoring and Adjustment

People often start judgment process with an initial value and alter evaluation around this anchor.

Poor initial values Insufficient modification from new information

Recency and Primacy Effects In remembering lists of items, memory for initial items

and final items is better than memory for “middle” items. Contributes to Anchoring and Availability Heuristic

Causal Schemas

Confidence in a conclusion is higher if you can construct a causal scenario that leads from one situation to the other and is in line with one’s expectations

Which prediction would be more accurate? Predicting a boy’s height from his father’s height Predicting a father’s height from his son’s height

Also occurs in jury decision making Judgments of guilt and innocence are often based on

juror’s ease of constructing a coherent story from the evidence.

Models that account for JDM process

Economic models predicted rational choices Obey the law of contradiction

People’s choices are not always optimal in their decision making

That does not mean choices are bad Psychologists have set up particular circumstances in

which people make poor choices Helps to illustrate processes people use.

Models of choice behavior Many different processes are used to make choices

Prospect Theory

Similar to economic models Value = Σ (π*ui)

π is subjective probability u is the utility of each option

Utility is evaluated relative to a reference point rather than some absolute utility

Accounts for framing effects

No objective probability, but rather subjective probability Objective probability is weighted by various

psychological factors

Prospect Theory

p is the objective probability of certain outcomes resulting from a given choice, π the subjective weight given to such probability

Much more subjective weight given to higher objective probabilities, much less to lower ones (“not well-behaved at endpoints”) E.g. weight given to p

= .9 not just the combined weight of say, .6 and .3

Regret Theory

People may make choices to avoid regret See Arthur Schopenhauer (19th century)

Status quo bias People would prefer not to make a change If a change is made, and it goes badly, there is

regret Regret is often overestimated

Reason-based choice

Imagine that you have just taken a tough qualifying examination. You feel tired and run-down, and you are not sure that you passed the exam. In case you failed you have to take the exam again in a couple of months. You now have an opportunity to buy a very attractive 5-day vacation package in Hawaii at an exceptionally low price. The special offer expires tomorrow, while the exam grade will not be available until the following day.

Do you sign up for the trip? Pay a non-refundable $5.00 deposit to decide on purchase after learning the results the next day? Not buy it?

A majority of people given this scenario pay the $5.00. (about 35%, 60%, 5% relatively for the options)

Two other groups are run One group told they passed the exam:

Most choose to go One group told they failed the exam:

Most choose to go

Reason-based choice

People want to be able to justify their choices May make decisions that are easiest to justify Shafir, Simonson, & Tversky

People want a reason to go on the trip. If they get a passing grade: Celebration If they get a failing grade: Consolation

Other reason-based effects The attraction effect

Effect is stronger if people have to justify their choice Justification is not always good

People tend to use less information and to rely on single dimensions when forced to justify a choice

It is easier to come up with these simpler justifications

Effort Accuracy framework

Dealing with complexity People attempt to make accurate choices

People want to minimize effort Some methods for making choices are highly

accurate Involve considering a lot of information Calculating expected utility is a high effort-high

accuracy way of making a choice. Some methods are simpler

Involve considering less information

Satisficing

Simon Choose the first option that is satisfactory

Will find an option that satisfies the goal Does not guarantee finding the best option

Imagine you are a manager at a supermarket You need someone to bag groceries You get 100 applications The cost of hiring a sub-optimal person is low Take the first person who looks like they can

do the job.

Elimination by Aspects

Tversky Start with the most important attribute

Eliminate all options that are not satisfactory with respect to that attribute

Then go to the next most important attribute Repeat this process until there is one option left

Lexicographic Semiorder Like Elimination By Aspects Look at the most important attribute Select the option that has the best value on that

attribute

Mental Accounting

Thaler Utility theory is a common currency theory

All options are evaluated with respect to utility But all gains and losses are not viewed as the same

People seem to have a variety of mental accounts

Imagine you are shopping for a calculator and a jacket, and you find them both at the same department store. The calculator costs $25, and the jacket costs $120. You are told that a store across town has both items, but the calculator is $15 cheaper at that store. Do you go across town?

Most people say yes. If the jacket is $15 cheaper, most people say no. In each case, they have spent the same amount of money.

Mental Accounting

The idea is that people are creating separate mental accounts for different goals. Money for necessities Money for entertainment Spending money from one account does not affect others

Imagine you have gone to the movies to see a show. You got to the front of the line and realized you lost $10, do you still go to the movie? Most people say yes

Imagine you have gone to the movies to see a show. The ticket costs $10. You buy the ticket early in the day. When you get to the theater, you realize you lost the ticket. Do you buy another one? Most people say no

Mental Accounting House money effect You go to a casino and put a quarter in a slot machine. You win

$100. How is your gambling behavior affected?

People are often more willing to gamble in this situation

Not any windfall increase in money works. You are just about to go into a casino, when you see a

newspaper. You own 100 shares of a stock and find out that it went up $1.00 that day. How is your gambling behavior affected?

Most people’s gambling behavior is unaffected by this news

In the first case, people feel as if they are gambling with the house’s money. In the second case, it feels like their own money.

Adaptive Decision Making

People adjust decision-making strategies in an adaptive manner

Satisficing, elimination by aspects, utility, random choice may all be utilized depending on the situation

Payne Little time pressure, complexity normative

decision making procedures More pressure, complexity more reliance

on heuristics

What makes a good Decision Maker? Use the best sources of information possible Base decisions strictly on the information

given