Exam 2

Question 1

Blink and The Power of Intuition Readings

Answer 1

Blink: thin-slicing is the ability of our unconscious to find patterns in situations and behaviors based on very narrow slices of experience. Works well. For example, marriage counselors can see what will happen from quick conversation between husband and wife. Dorm room shows stuff about the person’s personality that you wouldn’t necessarily know otherwise. Good predictor better than other things.

The Power of Intutition: intuition is relying on experience to make decisions. Extrasensory perception (ESP): people use this to make decisions. Experiences lead to intuition
Ex: nurse with baby example where she knows what causes the problems.
Ex 2: navy officer shoots down a plane, just know it is a missile not a plane and can just tell this is the case
Ex 3: firefighter: leave living room because not right, not sure what, just know there is a problem. The fire winds up being in the basement. Only knows this because of his ESP and intutition.

Question 2

Cognitive reflection test

Answer 2

Test designed to get you to say your gut answer, which winds up being wrong. Ask a question and require a quick response. The intuitive answer is wrong, though.
Ex: a bat and ball cost $1.10 in total and the bat costs a dollar more than the ball. How much does the ball cost?
Intuitive, gut answer is 10 cents. But when you think about it you realize it is 5 cents.

Question 3

System 1 vs system 2

Answer 3

System 1: intuitive system
It is fast/automatic, emotional, grounded in association/personal experience, simplifying
System 1 is the feeling mind. Intuition: a feeling or belief with an unknown origin. System 1 is a feeling of knowing or a feeling of preference. This feeling serves as information that we use to guide our judgements and decisions. Answer what you should choose or do by thinking about how you feel about it.

System 2: deliberative system.
Slow, cognitive/deliberative, abstract knowledge matters, more complete/integrating.

In general, system 1 starts and then system 2 kicks in.

Question 4

How system 1 and 2 interact

Answer 4

Approval
Solo operation
Neglect
Override
Inform
Influence

Question 5

Approval

Answer 5

Problem that requires an evaluation, assessment, judgement, or decision goes into system 1.
System 1 responds: “I’ve got a feeling”
System 2 responds: “looks good”
And you go with what system 1 said.
Response: evaluation, assessment, judgement, or decision

Approval in that system 2 approves the system 1 response.

Question 6

Solo operation

Answer 6

System 1 provides no response, system 2 responds.
There is a problem that requires an evaluation, assessment, judgement, or decision.
System 1 is unable to do it (for example, problem is 24*17. System 1 says “nope, I’ve got nothing”)
System 2 takes it.
And system 2 responds with the evaluation, assessment, judgement or decision.
—> system 1 provides no response and system 2 responds instead.

Question 7

Neglect

Answer 7

System 2 allocated insufficient resources to overrule system 1.
The problem goes to system 1
System 1 has a feeling
System 2 says “whatever, I can’t deal with this right now”
And you go with the system 1 response (evaluation, assessment, judgement, or decision)
We are most likely to go with our intuitions when:
1. We are rushed
2. We are tired (mentally or physically)
3. We are not paying close attention

We see that here in this neglect case, as system 2 neglects to respond and system 1 takes control. Using your intutition.

Question 8

Override

Answer 8

System 2 overrides system 1 response
The problem that requires an evaluation, assessment, judgement, or decision goes to system 1.
System 1 has a feeling.
System 2 overrides, and says “whoa, no. Let me do this instead”
Then system 2 has the ultimate response with an evaluation, assessment judgement, or decision.

Point here is system 2 gets to override the system 1 response
For example, if you really want to eat dessert, system 2 gets to override this sensation and tells you to not eat the dessert.

Question 9

Inform

Answer 9

System 2 receives information from system 1 as an input to a judgement or decision
The problem requiring an evaluation, assessment, judgement, or decision goes to system 1. System 1 has a feeling.
Goes to system 2. System 2 responds “thanks, that was very helpful”
System 2 then makes the decision based on what it learned from system 1 (response)

Question 10

First impressions

Answer 10

Thin slices do a good job of informing us about someone. But some stereotyping can occur, resulting in errors.
Generally, you can learn a lot about someone just from the first impression. Can go to their dorm room, for example, and learn a lot about someone.
You can distinguish different people just by looking at them.
We see this with voting decisions with the more competent looking candidate winning more often. See this in various elections.
Get first impression of 2 shapes, one named Takate and one named Maluma and you have a feeling Takete refers to the one with sharper edges. Just sounds like it. You have a feeling of it.

Question 11

Big and bad (SUVs)

Answer 11

People tend to think SUVs are more safe
But, you tend to only think of passive safety (I.e. if you are hit)
You need to consider active safety as well, which is what happens when you’re driving and avoid accidents.
We see it is less safe in that case
First impression/intutition can fail you sometimes.

Question 12

System 1 and 2 interaction: influence

Answer 12

When system 2 tries to override system 1 but is still influenced by it.
There’s a problem that requires an evaluation, assessment, judgement, or decision
System 1 responds “I’ve got a feeling”
System 2 responds “I know I shouldn’t listen to you, but…”
And what happens is system 2 makes the decision and is influenced by system 1 even though trying not to be influenced.

Question 13

The Feeling Mind

Answer 13

People often answer the question of what you should choose or do by answering how you feel about it
We see this with choosing between 2 bowls. You get $1 if you select a red bean
Bowl 1 has 10 beans and 1 red. Bowl 2 has 100 beans and 9 reds.
Even though 1 has higher probability of red, you likely select 2. This is because you just have a feeling of there being a higher chance of getting red in bowl 2.
You know it’s stupid but feels better to do bowl 2.
We see something like this when you read different words that are colors and they are in different colors. Confusing what to do.

Question 14

How do intutition influence us?

Medical decision example

Answer 14

Participants imagined suffering from angina (a painful but not life-threatening heart condition) and that the medicines they’d be taking have not relieved the pain
Participants chose between 2 treatments which they were told could potentially cure them of their angina: balloon angioplasty or bypass surgery

If told:
bypass surgery has one week in hospital and 75% success rate
compared to overnight stay in hospital and 50% success rate for balloon angioplasty
27% choose bypass surgery. Don’t want the week in the hospital

If then present this information is with 100 people and coloring each with different colors depending if likely to survive or not (75 people for bypass and 50 for balloon), then more people choose bypass (40%)

If give more anecdotes in favor of bypass surgery, more likely to do it.

Question 15

Feelings vs other information

Answer 15

Feeling usually and more compelling than other information
Conjecture: emojis elicit different feelings rather than plain text. Stronger emotions in response, highlights it better than just text.

People think differently about a question depending on how it is phrased, see this with the Bush bailout plan vs just saying it is trying to keep financial institutions secure.

See with abortion as well: “I believe in a woman’s right to choose” vs “I don’t believe the government should tell people when they should start their families”

And Obamacare vs affordable care act. Same thing but different phrasing.

It matters what you name things. For example, plastic leather vs vegan leather tote.

Question 16

When is intuition good

Answer 16

When people have repeated experience in a domain that provides near-perfect (precise/unchanging), immediate, and unbiased feedback.

Question 17

Why intutition often goes bad

Answer 17

Emotional attributes are misleading (attractive job candidates)
Over-application of available information (diagnoses based on recency)
Learned associations are overgeneralized (stereotypes)
Bad feedback (lie detection, complex business decisions)
Biased feedback, motivated reasoning (political experts, sports gamblers)

Intuition is similar to overconfidence in that more confident when you have expertise.

Question 18

Your Faulty Gut
Noise
Who’s on First readings

Answer 18

Your faulty gut: don’t always rely on intuition. Instead, use data. Ex: we think nba players are poor and from single parent families but data shows opposite.

Noise:
Bias is when there is a cluster around the wrong value
Noise is when you are dispersed around the right value, lot of differences caused by variabilities of judgement. Use algorithms to lower noise and standardize

Who’s on First: go against the book/intutition and use evidence. See this with money ball, for example, value the walk when other people focusing on average, for example.

Question 19

Alternatives to intuitive decision making

Answer 19

1. Experimentation/pilot testing
2. Base rates/find similar cases
3. Statistical models/consistently-applied decision rules
4. Aggregating (independent) opinions.

Question 20

Admission decisions: human vs algorithm

Answer 20

Human method: a judge decides which person to admit
A person uses information to make predictions

Statistical method: use past data (regression analysis/algorithms) to figure out which weights you should assign to GPA and SAT to best predict first year undergraduate performance. (Generalizing this, using past data to figure out which weights you should assign to different pieces of information to make predictions).

Statistical method is better than human method in general
We see this with disease severity and longevity correlation and heart attack prediction better with algorithms.
Hard for person/doctor to apply consistent decision rules while algorithm can do so easily.
Money ball: see this in action with statistical method proving better than humans.

Question 21

Why is statistical judgement consistently better than human judgement?

Answer 21

1. Human judgement is unreliable
   —> A. reliability: how much do two predictions of the same thing correlate with each other?
   —> B. Validity: how much does our prediction correlate with the outcome we are trying to predict?
   Difficult to have validity without reliability.
   We also see that experts are unreliable generally and have a tough time doing well on multiple attempts (low correlation between success rates in time 1 and 2 - shows that may be lucky you got it right the first time)
2. Human judgement often incorporates useless information and/or fails to incorporate useful information
3. Human judgement is not great at optimizing

Question 22

Why is human judgement unreliable

Answer 22

Human judgement is susceptible to:

1. Incidental factors (recency, order of information, mood, etc)
2. Random error that leads to inconsistent weighting of criteria
3. Fatigue/distraction

Example is seen with MBA interview evaluations. Interviewers relying on how they rated the previous person to rate this person. Will misrate someone because gave everyone else the same score. I.e if everyone gets a 5 and this person is last, then even if past person deserves a 5, will get a 4.

Question 23

Relying on useless information

Answer 23

See this with the mba interviews where relying on your other interviews to grade the last person
Ex: Tall people more likely to be paid more, become CEOs

Ex: baseball scouts care a lot about running speed and athleticism though it very poorly predicts performance

Ex: many firms and organizations hire based on interviews, even though interviews don’t often predict job performance and lead to bias

Ex: doctors assess the seriousness of a patient’s chest pain often use demographics to help make this determination, but using demographics doesn’t improve diagnoses.

Ex: Parole decisions: can be subjective. Use algorithm to get rid of this

Ex: Visual appearance mastering when doing a musical appearance even though it shouldn’t.

—> We also see that humans fail to use useful information
Ex: many expert fans ignore home field advantage when predicting sporting events.
Ex: Baseball scouts ignore highly predictive stats when assessing player talent.

Question 24

Caveats of using algorithms

Answer 24

1. The data used to create the model must reflect the range of possibilities in the real world (I.e. need to include possibility you could break a record)
2. You must have a quantifiable criterion, using quantifiable attribute values.
   Statistical models can neglect “soft” qualitative attributes that, in some cases’ should matter. The solution is to quantify them

You may be hesitant to use algorithms for subjective things like relationships.

Question 25

Why do people choose humans over algorithms? | How can we get people to choose algorithms over humans?

Answer 25

Consider using Waze vs using your own intutition. If you listen to yourself and change the route and get stuck in traffic, then you’re still confident in your own judgement. But, if you use Waze and change the route and get stuck in traffic, then you really blame the GPS The algorithm has a lot less wiggle room. Less content with algorithms mistakes. Model performing bad, you blame model. If you perform bad, not your fault. Example with algorithm aversion is using your own judgement to make some sort of estimate of using the model to do so. 10 rounds. You either use model for all 10 of your own for all 10. Human forecasts have 15% more error than the model Yet, will choose people over model after seeing model perform. This is because you are less tolerant of algorithm mistakes.

Question 26

Overcoming algorithm aversion

Answer 26

In this experiment, offer 4 cases: can either not change the model. Or can change the model by 10%, 5%, or 2% People much more likely to use model when you can adjust it. But doesn’t make a difference if you can change by 2, 5, or 10 percent. People were insensitive to the amount they could adjust the algorithm. It seems people want to have some control over the algorithm, but not necessarily greater control People choose humans over algorithms because they are less tolerant of algorithms mistakes. We can get people to choose algorithms but letting them modify them slightly. Also to improve judgement: 1. Improve reliability: have independent indiviudal’s make the same judgement (larger sample size) 2. Discard useless information: blind the judge to useless info. One last caveat: algorithms are still made by people. Still can have human bias in them.

Question 27

The Wisdom of Crowds | Polarization: One Reason Groups Fail reading

Answer 27

1. Use groups to get better answer than if you did so individually. See how pooling together averages out good and bad and gets to a good answer. 2. Groups lead to polarization and more extreme views.

Question 28

3 methods to combine opinion

Answer 28

1. Group deliberation/discussion 2. Averaging opinions 3. Prediction markets

Question 29

1. Group deliberation/discussion

Answer 29

Many decisions are made in or by groups. There is a general belief that group deliberation represents a good way to make a good decision. But is it effective? Example is with Boulder liberals and Colorado Springs conservatives being in their own groups and being exposed to a statement about the US and greenhouse gases. Then asked to measure their attitude toward a global warming agreement. We see the group makes them more polarized, especially when with similar people or background. Groups tend to go more extreme than people originally. People with strong beliefs speak 1st and anchor the discussion. Group polarization: the tendency for a group to make decisions that are more extreme than the initial inclinations of its members. Hidden profiles: ex: asked to choose who to vote for. information strongly favors candidate A. Some groups, all members get all information about the 3 candidates. Some groups, each member received only a subset of info supporting candidate A. When you get all the information, lot more likely to support A and will like him even more after discussion. When partial info, less likely to support A and even less likely after discussion

Question 30

3 problems with discussion

Answer 30

1. The potential for bias: Ex: How can so many people be wrong, looks like everyone already agrees, my (minority) opinion won’t matter anyway 2. Some opinions aren’t voiced (or even invited): Ex: i am too afraid to speak up, or don’t care enough to speak up. Boss chooses only top employees of nearby employees 3. Opinions aren’t independent. In real groups, afraid to speak up. Also listening and not thinking. So, more and better ideas when you write down your ideas on your own instead.

Question 31

Ingredients for quality inputs | Make sure the group is:

Answer 31

1. Knowlegable 2. Diverse/unbiased 3. Independent 4. Motivated to contribute (all information is extracted)

Question 32

2nd method of combining opinions: averaging opinions

Answer 32

See example of the power of aggregation with Galton and the Ox, where in 1906, Galton asks people to guess the weight of the Ox. Off by 1 pound even though people don’t know it. Point is that averaging opinions is often a way to elicit good information from a group of people, as averaging cancels out idiosyncratic errors and biases.

Question 33

Measuring accuracy

Answer 33

Absolute deviation = (your guess - true answer). Average deviation is the average of the absolute deviation. So for everyone, your guesses - true answer. Then divide by total guesses. Deviation of the average = actual - average This shows how far off the average is to the actual value. % worse than average: the percent of guesses who do worse than average. See this tends to be high, as the average is a great predictor.

Question 34

Third method for combining opinion: prediction markets

Answer 34

Bet on an outcome happening. Ex: HP employees use prediction market to forecast sales of various products at least 3 months in advance. Independent of that, HP also made official forecasts. The prediction market outperforms their official forecasts 6 of 8 times. On average, prediction markets outperform by 4% On the whole, prediction markets are extremely accurate at predicting outcomes.

Question 35

Why prediction markets work and when do they not work?

Answer 35

They work bc: Knowledgeable, diverse/unbiased, indeed, and motivated to contribute (all information is extracted) They don’t work when: Prediction markets are susceptible to rumors, fads, biases, and bubbles. Prediction markets won’t work when the event’s occurrence is affected by the outcome of the market. For example, terrorist may only strike when market predicts they won’t.

Question 36

4 ingredients for quality inputs.

Answer 36

Make sure the group is: 1. Knowledgeable: think creatively about who has the knowledgeable information 2. Diverse/unbiased: cancel out idiosyncratic biases 3. Independent: elicit opportunities seperatly 4. Motivated to contribute (all info is extracted)

Question 37

Reversals | The Truth about Relativity readings

Answer 37

Reversals: view things next to each other to have a sense of comparison. Reversal of judgement when you do this. Seen with examples of bets, height, dolphins/farmers where you change your choice when you see two things next to each other. Truth about relativity: relativity in decisions: slightly worse version of same thing leads you to pick that thing instead of a tough to compare option. For example, if you choose between small and large popcorn, you will make your choice. But then if you add medium as an option that is close in price to large, you’re more likely to choose large.

Question 38

Choice overload

Answer 38

Choices are difficult. You don’t know what to choose. Choice overload is when there are so many choices that you don’t know what to do. Unclear how true this is in academic studies. There is a widely cited study showing when given 6 choices of jam vs 24 choices, you are more likely to buy when given the 6 options. But it isn’t replicated well. People often like to have fewer choices or go with the default. Too many choices is a problem: choice overload. Information overload is related, where if you give people too much information, they don’t like that. Decision fatigue: too many decisions results in you being tired. Go with the default.

Question 39

Choice blindness

Answer 39

You don’t know your choice. Experiment done where you choose between 2 people’s faces. And you select one and they give you the other one and you think it is the one you selected and are able to justify the wrong one! Rationalizing the choice you didn’t choose! When you don’t have strong preferences, can justify either choice and don’t even remember choice.

Question 40

Constructed preferences | Example with valuing future lives

Answer 40

Valuing future lives: 1. Program A will save 100 lives now vs program B saving 7000 lives 100 years from now. Usually you are indifferent here. This implies 1 life today is worth 70 lives in 100 years 2. Program A saves 300 lives in your generation and 0 after. Program B saves 100 for 3 generations. Here, 80% choose B. This implies a typical person thinks 1 life today is worth 1 life in 100 years. 3. Program A saves 100 lives this decade, 200 next, 300 in three decades. B is reverse. Here, 71% of people choose A. This shows a typical person thinks a life is worth less than a future life Ultimately, this shows us how while we think we have stable, ordered preferences, we actually construct preferences on the spot. We don’t have a set conversion rate and it depends how we phrase things. Depending on how it is framed makes your choice. Constructed preferences: When choices are not well rehearsed, we often construct our preferences on the spot. We search for reasons for choosing 1 option over another. We use the way an option is presented as information that helps us construct our preferences.

Question 41

Sure-thing principle

Answer 41

If you prefer A over B under condition X and condition not-X, then you should prefer A over B when you don’t know whether X or not-X will occur This means, if you prefer A over B with either outcome, you should prefer A over B when you don’t know the outcome. Ex: person wants to buy property but he is waiting until after the election. He decides he will buy if a republican wins. He decides he will buy if a Democrat wins. So, since he will buy in either case, he should just buy it now. Another example is with a vacation and an exam, and if you know you’ll pass, you buy it and if you know you’ll fail you’ll buy it. So may as well buy if you’re uncertain as well. Third example: if there is a second gamble that you would take if you won or lost the first one, you may not want to take it if you are uncertain even though you would take it in either outcome. —> we don’t like making choices when there is uncertainty even when the uncertainty shouldn’t matter!

Question 42

Evaluability

Answer 42

Some numbers need a context in order for us to know what they mean. For example, is a 4 mm diamond small or large. Is a program saving 1000 lives large or small? Example is seen when paying for ice cream, when you have a small cup overflowing vs a large cup not filled, you will like the small cup better on its own, but when together, you vslue the large cup higher because it is larger (even though not filled). See same thing if you have a program that saves 80% of a forest of 5000 birds and another that saves 20% of 25,000 birds. You would think the one with the higher percentage would be better when evaluated seperatly. But together, you prefer the 20% as more total lives. When presented in isolation, people will be more inclined to favor a program that saves a higher proportion of the victims than a program that saves a lower proportion but more lives total.

Question 43

Unit asking

Answer 43

People aren’t as sensitive to scale as they should be. For example, a school principal asks for donations to buy Christmas gifts. If asking for donation for 20 kids, you get $18 If ask for donation for 1 kid, you get $15 Then if you ask that same person who gave $15 how much they would give for 20 kids, you get $49 So, we see that people aren’t as sensitive to scale as they should be. Asking for 1 kid and then 20 gets you more for the 20 kids than just asking for 20 right out of the gate.

Question 44

Opportunity cost neglect, naive diversification, and allocation decisions

Answer 44

Opportunity cost neglect: people forget what else they can do with money Ex: you’re at a video store and you see a new video worth $14.99. If asking if you would buy or not, most people would buy. But, if ask if you would buy or “keep the $14.99 for other purchases,” lot less people buy. Naive diversification: people often evenly allocate between choices when not sure what to do. We see this with stocks vs bonds allocation questions, people usually do 50-50 Will do 50-50 splits between funds when not sure what to do. Depending on how the choices are presented, you make different choices. If you split them out, then more likely to choose between. For example, if choosing between US and international, may split 50-50. But if it is between 4 cities in US and international, then less goes to international. We see a similar thing on the menu when ordering tapas. More likely to choose something when it is separated out. —>Choose something if it is unpacked rather than packed together.

Question 45

Lesson from how we construct preferences

Answer 45

Traditionally: we have stable, well-ordered preferences Reality: we often construct our preferences on the spot and are influenced by how choices are presented.

Question 46

Making reasons salient

Answer 46

This is about putting an important reason at the forefront. Berkeley business students were offered a choice between attending MBA programs at northwestern and UCLA All participants exposed to response of a previous participant who selected northwestern. Either given no reason or say they do so because they have many relatives in Chicago We see that those who are exposed to the condition of “I have many relatives” are way less likely to choose northwestern than those who are exposed to no reason. Point of this is if you are exposed to bad reasons and bad reasons become salient, they can hurt. Example seen with bad giveaways. People assume all giveaways are positive or neutral, but they can actually be negative. See this with the pillsbufy brand cake mix

Question 47

Assumption vs reality number 2

Answer 47

Traditionally: providing reasons for choosing an option should have a positive (or neutral) effect on choosing that option Reality: making bad or irrelevant reasons for choosing an option salient can have a negative effect on choosing that option.

Question 48

Adding options and the lesson from that | Asymmetric dominance and compromise effect

Answer 48

Traditionally assume that adding options shouldn’t add to the choice share of any existing option. Instead, in reality we see that adding options can give a reason to pick an existing option, adding to the choice share of the option. (Assymetric dominance and compromise effect) If people prefer A over B, then people should prefer A over B even in the presence of a third option, C But, adding an additional option can create a context that offers a seemingly good reason for choosing that wasn’t there when the option was not presented Ex: economist subscription. If being just offered an online subscription for 60 and online + print for 125, most people will likely choose online But, if you add an option for only print for 125, then the print + online subscription looks a lot better. People will choose the print + online because you added this other option. Asymmetric dominance/decoy theory: People tend to have a change in preference between 2 options when presented with a third option that is asymmetrically dominated. We see this with the economist subscriptions, as the print only option makes you want to do print and online. Compromise effect: people tend to avoid extreme choices (which seem risky) and instead tend towards middle options (which seem safer) Providing a more expensive option makes the other one better We see this, for example, if you’re selling TVs and want to sell one particular one, put it with one more expensive and one cheaper and sell that middle one.

Question 49

Common attributes

Answer 49

Showing stuff that different options are the same at can help improve information and choosing. Helps you gain a better understanding of the different options.

Question 50

Procedure invariance and lesson from this

Answer 50

Traditionally: logically identical methods of preference elicitation should be logically consistent Reality: people sometimes respond differently to logically identical methods of preference elicitation. Procedure invariance: preferences over prospects should hold regardless of how you ask people about those preferences (choice, willingness-to-pay, ratings) But, people sometimes respond differently depending on how it is phrased. Ex: people will pay more for a high value bet. If asked between a 28/36 chance to win $10 (high probability) or 3/36 chance to win $100 (high value), more people choose high value.

Question 54

Prospect Theory when Six of One isn’t half a dozen Readings

Answer 54

Prospect theory: based on relative choice, change in wealth, loss aversion, S shaped utility curve with more negative steep slope (loss averse) Problems: reference point could change (could be mad to get nothing if high chance of winning something), could regret something and regret isn’t in prospect theory Six isn’t one half dozen: with prospect theory, we are overly averse to loss and have high sunk-cost bias. Can lead to bad decisions: don’t sell stock because losing money. Sunk cost-don’t go to bulls game if it’s snowing and you got free tickets, but go if you bought the tix.

Question 55

Prospect theory and how utility function differs from traditional utility function

Answer 55

Prospects are evaluated based on expected utility. Traditional utility theory: the “carriers of utility” are states of wealth, not gains and losses. We plot utility on the Y axis and income or consumption on the X axis. Only graph the top right quadrant. We see an increasing function, but gets less steep as goes on, as there is diminishing marginal utility. Prospect theory value function: we now graph utility on Y axis and losses and gains on the X axis (not income) This graph shows all 4 quadrants and starting in the bottom left is flat, then goes very steep up to the reference point (0,0). Then goes up steep (but not as steep as the negative steep before that) and then gets less steep as goes on. So, there is diminishing marginal utility and steeper loss curve. 1. Steeper loss curve: loss aversion 2. Plotting gains vs losses (reference point at 0) rather than income 3. Diminishing sensitivity to gains and losses (less steep as get more loss or more gain) —> people are sensitive to gains and losses, rather than to final states. Gains are good. Reference point is fine. Losses are bad.

Question 56

Prospect theory risk preferences

Answer 56

We don’t want to give up sure gains, and so we are risk-averse in the domain of gains. We don’t want to accept sure losses, and so we are risk-seeking in the domain of losses. Ex: if given $1000 and choosing between sure gain of $500 and 50% chance of gaining $1000, you will almost always take the sure gain But, say we’re given $2000, and you can choose between a 50% chance of a loss of $1000 and a sure loss of $500, we see that most people will choose the 50% loss. This is because people are loss averse. They don’t want the sure loss, but want to make that gamble. But, they will take the sure gain. More risk seeking in the domain of losses. We see that rephrasing the question can change the outcome. When dealing with gains vs loses, you completely change your preferences. Ex: outbreak of disease expected to kill 6000 people. Can either save 2000 or have 1/3 probability save all of 2/3 prob all did. You choose the save the 2,000 people for sure. But, reframe situation. Imagine disease expected to kill 6000 people and one program will have exactly 4,000 people die and the other will have 1/3 prob nobody dies and 2/3 prob all 6,000 die. Here, you would rather take the risk. In domain of losses, people are more risky.

Question 57

Reference points and prospect theory related to this

Answer 57

If people are sensitive to gains and losses, then they are sensitive to (even arbitrary) reference points. Ex: northwest airlines charged $10 extra for purchasing tickets offline while jet blue gave a $10 discount for purchasing tickets online. We see that people prefer to be rewarded for changing behavior rather than punishing existing behavior. Ex 2: with car insurance. If no accidents means reward and an accident is fine vs no accidents if fine and an accident is a penalty, you prefer the second one, as avoiding losses.

Question 58

Salient reference points

Answer 58

Some salient reference points: 1. Expectations/what’s typical - reference your gain/loss based on this Ex: valuing water, you value it differently if you’re getting it from a run-down grocery store vs a fancy resort hotel even if it is same water and asking just about your value. True value changes based on expectations of deal We see this with transaction utility: people are sensitive to the value of the deal. We buy some things that we don’t really want primarily because they are a good deal and we fail to buy things that we do really want because they are a bad deal. 2. Status quo - reference based on the status quo - compare to what you had Because the status quo often serves as the reference point, people are sensitive to deviations from the status quo For example, raising taxes is a loss but failing to cut taxes is no loss. And cutting salaries is a loss but failing to increase salaries is not a loss 3. What could have been - reference based on the thing you had and didn’t get Ex: Mr. A is waiting on line and he is the 100,000th customer so he wins $100. Mr. B is on line at a different theatre and man in front of him wins $1000 for being the 1,000,000th customer. Mr. B wins $150 Mr. A will be happier than B even though B won more money. This is because you’re thinking about what could have been, and B could have won 1k This is seen with silver medalists not being happy as thinking about winning first (bronze happier than silver), with deal or no deal when you think you could have won a lot but took a deal early, and with last code lottery if the zip code next to you wins or if the office does a lottery and you don’t participate and you win. There is some outcome bias with all of these, especially the deal or no deal.

Question 59

Lessons from reference points

Answer 59

Actively shift your own reference points Adopt the most appropriate reference point (often, total wealth) Know what others’ reference points will be. Since looking at gains and losses relative to reference point, the reference point is important. We see an example of this with saving $10 off a $15 calculator of off a $125 coat. We value the $10 greater with the calculator since reference point to the price and is larger portion of it. Another example with when you buy a car, you prefer to spend the extra $200 for warranty on the 25k purchase during the purchase than later, as the separate transaction is annoying. More likely to buy warranty when offered it during the transaction because of this. Although, when getting a gain, prefer two payments of 50 than one of 100.

Question 60

Combining gains and losses

Answer 60

Consider two health club managers: Manager A’s customers receive a monthly bill ($40) and manager B’s receive a monthly charge ($40) to their credit card. Manager B is happier, as not physically paying it. Only see the charge once not twice (see and pay for A). Consider customers getting a discount when buying a car vs getting the discount a week later. Prefer to get the discount a week later. And with rebates, prefer to get a 20-cash rebate rather than reduce the price. Rebates are effective because at the time of purchase they are seen as a separate gain, instead of as a reduction of a loss. We see that people want to combine losses but separate gains. Want to get the gain twice. Want to lump the loss in. With diminishing sensitivity, lesson is that we actively shift contests when deciding how much you value something. We integrate losses and separate gains.

Question 61

Bitter Money and Christmas Clubs Knowing When to Pull the Plus Readings

Answer 61

1. Not as strong as we wish to be. Mental accounting is treating different types of money differently - placing them in different accounts Ex: x-mas club. Give back money on an interest free loan so have money for the holidays. Not rational but do it anyway. 2. Managers overcommit to projects, scared to abandon them. Over commitment and overconfidence. Organizational and social pressures are involved in this.

Question 62

Loss aversion

Answer 62

Negatives/losses matter more than positives/gains People say no to even positive expected values, as you are scared of losing. Ex: most people wouldn’t play a gamble that offers a 50% chance of winning $10 and 50% of losing $10 Ex 2: most people wouldn’t be willing to do a gamble that offers 50% chance of winning $25 and 50% of losing $22 Ex 2B: if you add wealth to this and assume your wealth is 1,200 and then have that a 50% chance of having $1225 and 50% chance of $1178, we think of it differently as we frame it differently. Losses hurt more than equal gains feel good. Ex 3: even though you like cake, you hate a cockroach so much that you wouldn’t eat a cake w a cockroach on it Ex 4: would require more money if you participate in a study and that leads you to develop a disease than if you just got the disease.

Question 63

The endowment effect

Answer 63

Sellers prices are greater than buyers prices Ex: sellers were endowed with mugs. And buyers weren’t endowed with mugs. Sellers pricing it at $5.25 and buyers at $2.25. Massive difference. Endowment effect: on average, selling prices tend to be about 2x greater than buying prices. Giving up something is roughly 2x more painful than getting something is pleasurable. This is partly because of loss aversion - don’t want to give up (sell) something. Ex: Seinfeld doesn’t get his rye bread, willing to pay a ton for it, but the lady who gets the last one doesn’t want to sell it because she values it so high. Jerry offers her $50 and she declines even though she never would have paid that originally. Sellers overvalue. Can use this effect to increase sales by giving a free sample/free trial (of a renewing thing) or free returns. That way the customer already has it and it feels like a loss when you take it away. Leads for it to be valued higher in that situation.

Question 64

Endowed progress effect, ikea effect, golf performance, and marathon runners

Answer 64

Endowed progress effect: Example: look at car wash reward program. 8 car washes and get 1 free vs need 10 but already have 2 filled out for you for free. Lot more redeem free cash wash in the second case. Already have the progress, don’t want to lose it. Part of loss aversion IKEA effect: think item is worth more because you built it. Golf performance: closer to hole on par putt as loss averse, scared of bogey Marathon runners: use round numbers as reference points, want to beat them. For example, significantly prefer 3 hr 59 min than 4 hr 1 min. See a lot of clusters leading up to major round numbers.

Question 65

Implications of loss aversion

Answer 65

1. Disadvantages relative to a reference point are given greater weight than advantages relative to a reference point. 2. Negotiators value what they give more than what they get. 3. Price increases often have a greater effect than price decreases.

Question 66

Insurance and the lottery and what this means with probability weighting

Answer 66

If people are more risk averse with gains and more risk seeking with loses, then A. why do people buy insurance? B. Why do people play the lottery? A. Decision to purchase insurance is between, for example, -$500 for sure or a 0.01 chance of -$4k B. Decision for lottery is between losing $1 for sure or a .0000001 chance of $5M The reason for this is probability weighting. We overweigh very small probabilities and we underweigh larger probabilities. We think really small probabilities more likely to occur than not. And we think large probabilities less likely to occur than actually are.

Question 67

Certainty effect

Answer 67

People overweigh outcomes that are certain relative to those that are probable Ex: Russian roulette. Pay some amount to remove 1 bullet from the gun which holds 6 total bullets. Would you pay more to reduce the bullets from 4 to 3 or from 1 to 0? Typically, 1 to 0, as lowering the probability from some to none. Reducing the probability by 1/6 has a greater impact when it moves to or from certain outcomes like it does when prob goes to 0 here. Ex: imagine flood insurance which covers all losses due to flood and property insurance which covers all losses due to flood and 50% of losses due to fire. In isolation, the flood insurance is better, as it feels more certain. We prefer vaccines that are 100% effective for 70% of strains of cancer than those 70% effective for 100% of strains, as we value certainty.

Question 68

When do we overweigh small probabilities and when do we discount them?

Answer 68

Availability bias leads us to thinking something is more prevalent than it is. Overweigh the small probabilities when it is in the media. Example is shark killing you vs vending machine. Sharks are in media and feel scarier than vending machine. Whenever possible, give people certainty. People are relatively insensitive to shifts in uncertain probabilities.

Question 69

Prospect theory summary

Answer 69

1. Gains and losses: focusing on how you did relative to a reference point rather than how you did absolutely 2. Diminishing sensitivity: care less as you go on. Care more near the reference point 3. Loss aversion: want to avoid losses 4. Probability weighting/certainty effect: prefer certain things. Relatively insensitive to shifts in uncertain probabilities but like certain things.

Question 70

Mental accounting

Answer 70

Money has labels. Those labels affect spending, saving, outcome evaluation, and decision making. Ex: you receive $100 cash for Christmas and waiting to spend it for something special and still have that cash in your wallet months later although you’ve withdrawn and spent $100 many times since Ex 2: decline to buy sweater for yourself but then get as a present and you’re happy even though from your wife with joint bank accounts Ex 3. You prefer to receive $100 cash than a $100 gift card. But, you prefer to give the gift card. When you gamble, you feel like you didn’t have the money originally and treat if differently. More willing to spend it rather than something like salary which you want to save up. And cash in hand vs poker chip very different. Paper vs a realized loss. Treated differently. All of this is mental accounting, that money have labels and the labels affect spending, saving, outcome evaluation, and decision making.

Question 71

2 types of mental accounts

Answer 71

1. Topical boundaries This is largely related to justifying costs that have already been spent. Ex: movie tickets. You lose the ticket when you walk in, would you pay $20 for a new one. 40% say no. But if you lose $20 when you walk in, still would pay $20 for the ticket. Only 21% say no. We see how we treat money differently, and losing the ticket makes it seem like spending $40 on the play bc money was tagged for this while in the 2nd case the money is different Ex 2:more likely to ski on the 4th day of a trip if you bought 4 individual tickets than if one ticket for all the days, as it feels like you are wasting an entire ticket if you don’t. Ex 3: get compensated for something you lose and then spend that money like it is extra not on the thing you lost. 2. Temporal boundaries. See next lecture. Mostly based on timing things. We see this it’s investing and myopic loss aversion. See next lecture.

Question 72

2 lessons from mental accounting

Answer 72

1. People don’t treat all money the same, but you should 2. People are enticed by promotions that erase hated payments (more than equivalent promotions) Ex: people have paying for taxes. They prefer a discount that gets rid of taxes rather than an equivalent one just on the price. You will drive to avoid paying for something you hate, like driving to jersey for gas, for example.

Question 73

When Doctors Make Mistakes You Need Hands Readings

Answer 73

1. Doctors make mistakes, but rather than blaming the doctor, blame the process. Mistakes will happen, even to the best doctors. Try to prevent the mistakes and be careful. But establish a better process to help do this. 2. People only have 2 hands. Make shopping experience play into this. Put baskets to hold things throughout the store. Have shelves at right height to store things when shopping.

Question 74

Sunk costs

Answer 74

Lesson: people honor sunk costs, but you shouldn’t Sunk costs relate to when you already paid for something, but you follow through and do it even though the cost is already gone. Ex: tennis elbow - a man joins a tennis club and pays $300 yearly membership fee. After two weeks, he develops a tennis elbow. He keeps playing in pain as he doesn’t want to waste his $300. Wants to get his moneys worth. See a similar thing when Monica buys boots she hates but wears them bc wants to justify the purpose. Another example is with theatre attendance. With a discounted ticket price, you are more likely to not attend, as with a full ticket price you feel like you lose more, and have that account in the red. Feel like you paid for nothing. If you go you feel like you got what you paid for. —> if you pay $15 and don’t go, the theater account is in the red. Feel like you paid for nothing. But if you pay the $15 and do go, then account isn’t in the red. You got what you paid for. We see that people make an emotional balance sheet with how they feel about things. People buy extra things on amazon so get your moneys worth. Sunk cost fallacy: feel that if you spend something, you have to finish/use it. Should treat the money/time as sunk and move on. Ex: if you spend time on one essay topic, you feel you shouldn’t switch to another topic bc already used the time.

Question 75

Examples of temporal boundaries with investing and gambling

Answer 75

Temporal boundaries are mental accounts having to do with time Ex 1: imagine a 25% chance of winning $30 or a 20% chance of winning $45. Most people will do the second option Now imagine a two stage game. In the first stage, 75% of ending the game winning nothing and 25% chance to move to next stage. If reach the next stage, can have a sure win of $30 or an 80% to win $45. Here, most people choose the sure win of $30 —> what this shows is when you consider the second choice in isolation you are more risk averse, as it is essentially the same question. But we have different preferences depending on the time it is asked. Ex 2: imagine two investment options, stocks, which have a higher mean but large standard deviation and bonds, which have lower mean and no standard deviation. Imagine investor A who evaluates portfolio every month and B who evaluates every year. Investor A more likely to invest in bonds and B more in stocks. This is because lot more variability with stocks. And when you look monthly, potential for losses. Experience losses and gains more with monthly, more risk averse with monthly than yearly. More accepting of risk when over long period of time, as the larger sample size allows it to approach the true mean while any one time may have random outcomes.

Question 76

Myopic Loss Aversion

Answer 76

Consider a gamble where there is a 50% chance to win $110 and 50% to lose $100. If asked if you want to play the game once, you decline. If asked to play 5 times, you accept. Then ask if prefer 5 or 6 times, you choose 6. But then if you have already played 5, you likely wouldn’t do the 6th as you are loss averse. The lesson here is to take smart risks but don’t watch. Implications: because we often consider gambles in isolation, we avoid taking risks that we would take if we adopted a wider frame. Think of risky decisions in the context of all the decisions in your life. Take a portfolio approach and if you would play it 100 times, you should play it once if you can afford it. Ex: phone makes you worse investor as keep checking how you’re doing

Question 77

Sudden Death Aversion

Answer 77

Ex: prefer a game where you win money if you get a J, Q, K, or ace vs 2 rounds where you win if you get an 8,9,10,J,Q,K,A. People choose second game as prefer a longer death, even though first one has better odds. Sudden death aversion: people prefer slow strategies that reduce possibility of losing quickly compared to fast strategies that include possibility of immediate defeat even when the fast strategy provides a greater chance of success Ex: prefer going into overtime late in games than winning or losing right then. Main lesson: people engage in narrow framing but you shouldn’t.

Question 78

Prepayment vs post-payment

Answer 78

It is painful to pay for things after we are done consuming them. It is pleasurable to consume things after we have paid for them (we feel like we are consuming for free) Ex: you prefer to lose $100 in 6 months than now. You prefer to gain $100 now than in 6 months. Ex 2: washer and dryer machine. You prefer to pay monthly payments after receiving not before But, ex 3: you prefer to pay for the vacation before not after —> it is painful to pay for things after we are done consuming them. It is pleasurable to consume things after we have paid for them (we feel like we are consuming for free) Most people, though, like to get paid after they work because it makes the work more rewarding. We can mental prepay by setting aside money for something. De-couple payment from consumption To get people to save more: Use mental accounting to influence people to save more. Could do something like the Christmas club with savings. Lesson: people dislike when any account is in the red and so hate paying for something after consumption or working for something after being paid.

Question 79

The power of social influence: what causes evil?

Answer 79

Why did the holocaust happen? Popular layperson hypothesis: Germans are/were bad people Popular academic hypothesis: German culture is particular susceptible to social influences (conformity, pressures of authority, etc)

Question 80

The Milgrim Experiment

Answer 80

Participants told randomly assigned to either teacher or learner. Teacher delivered shocks to a learner for wrong responses. The shocks increased by 15 volts for each wrong answer. People in the experiment are teacher, learner is a confederate: participating on the inside but pretends to be a participant. When the experimenter would take responsibility, the teacher is more willing to shock. We see the teacher is willing to obey with the experiment and shock for a long time. Way more than expected. Milgrim blames the situation rather than the people. Lesson: Situations can have a powerful effect on behavior. Big effects (administering lethal shocks, abusing others, etc) can have small causes (guy in lab coat telling you what to do) The bad news: good people can do very bad things Good news: situations are easier to change than the people Need to consider contest of situation. Power of the situation in dictating how people behave. When doctors make mistakes: look at process rather than doctors. Change situation rather than the person. Important question isn’t how to keep bad physicians from harming, but rather how to keep good physicians from harming patients.