Incentives and Behavior Prof. Dr. Heiner Schumacher KU Leuven 9. - - PowerPoint PPT Presentation

Incentives and Behavior

• Prof. Dr. Heiner Schumacher

KU Leuven

• 9. Behavioral Biases I
• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

1 / 36

Introduction

Literature Kahneman, Daniel (2011): Thinking, Fast and Slow, Farrar, Straus and Giroux, New York, Part 2 (“Heuristics and Biases”).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

2 / 36

Introduction

In the last chapters, we discussed human preferences. “De gustibus non est disputandum.” Now we turn to behavioral biases. Behavioral biases (or decision errors) are behaviors that people would not exhibit if they would examine the situation with su¢cient care. From a welfare-economics perspective the existence of biases is

• problematic. If many individuals exhibit biased decision making, there

could be scope for public policy that helps them to make better decisions. Most of the material we consider in the following is due to the psychologists Daniel Kahneman and Amos Tversky.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

3 / 36

Introduction

Overview The Law of Small Numbers Anchors Availability Availability and Risk Representativeness

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

4 / 36

The Law of Small Numbers

The Law of Large Numbers. Let x1, x2, ... be an in…nite sequence

• f independent and identically distributed random variables with

expected value E(xi) = µ for all i. Then the sample average ¯ Xn = 1 n(x1 + ... + xn) converges to the expected value: ¯ Xn ! µ for n ! ∞.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

5 / 36

The Law of Small Numbers

Many people (scientists and non-scientists alike) exaggerate how likely it is that a small sample resembles the population from which it is drawn. This phenomenon is called the belief in the “law of small numbers”. The …rst study on this subject is Tversky and Kahneman (1971). They …nd that even mathematical psychologists greatly overestimate the likelihood that results that have been found signi…cant in small samples (e.g. 20 subjects) remain signi…cant when the sample size is doubled.1 The belief in the law of small numbers can lead to severe decision errors.

1Tversky, Amos, and Daniel Kahneman (1971): “Belief in the law of small numbers,”

Psychological Bulletin 76(2), 105 - 110.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

6 / 36

The Law of Small Numbers

As an illustration, take the sex of six babies born in sequence at a

• hospital. Obviously, the sequence of boys (B) and girls (G) is random

and the events are independent from each other. Consider the following three possible sequences: BBBGGG GGGGGG BGBBGB Are these sequences equally likely? Which sequence is the most likely

• ne?
• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

7 / 36

The Law of Small Numbers

Random processes produce many sequences that convince people that the process is not random after all. Consequently, people are very bad at producing random sequences. For example, Rapoport and Budescu (1997) asked their subjects to “imagine a sequence of 150 draws with replacement from a well-shu-ed deck, including …ve red and …ve black cards, and then call aloud the sequence of these binary draws”.2 The probability that a subject would produce a signal given that the previous 0, 1, 2, or 3+ signals chosen were that same signal were as follows:3 Pr(A j B) 58.5% Pr(A j AB) 46.0% Pr(A j AAB) 38.0% Pr(A j AAA...) 29.8%

2Rapoport, Amnon, and David V. Budescu (1997): “Randomization in individual

choice behavior,” Psychological Review 104(3), 603 - 617.

3These numbers are taken from Rabin (2002).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

8 / 36

The Law of Small Numbers

A phenomenon that is related to beliefs in the law of small numbers is the hot hand fallacy: If there is uncertainty about the data generating process, then many people treat even short random sequences as highly representative of the random process. Gilovich et al. (1985) show (irrational) beliefs in the hot hand in basketball.4 Players and fans alike tend to believe that a player’s chance of scoring are greater following a hit than following a miss on the previous shot. However, the correlation between outcomes of successive shots is zero. Camerer (1989) shows that the belief in the hot hand may have an in‡uence on betting odds.5

4Gilovich, Thomas, Robert Vallone, and Amos Tversky (1985): “The hot hand in

basketball: On the misperception of random sequences,” Cognitive Psychology 17(3), 295 - 314.

5Camerer, Colin (1989): “Does the basketball market believe in the hot hand?,”

American Economic Review 79(5), 1257 - 1261.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

9 / 36

The Law of Small Numbers

Do beliefs in the law of small numbers a¤ect important decisions? There is some evidence that the answer is yes. The Gates Foundation invested 1.7 billion USD in order to …nd out the characteristics of the most successful schools. The critical measures were math and reading skills in several grades. One conclusion from this research was that most successful schools,

• n average, are small. For example, in Pennsylvania, 6 of the top 50

were small (an overrepresentation by a factor of 4). This encouraged the Gates Foundation (and a large number of other prominent organizations) to invest a lot of money in order to split up large schools in smaller units.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

10 / 36

The Law of Small Numbers

It is easy to construct a story that explains why small schools perform better than larger ones. However, if one asks about the characteristics of the worst schools,

• ne also …nds a overrepresentation of small schools (by the same

factor). Small schools are not better than large schools; the test scores are just more variable.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

11 / 36

Anchors

Consider the following experiment. In the …rst stage, the subject spins a wheel of fortune that may stop either at 10 or at 65. She then has to write down this number. In the second stage, the subject has two answer the following two questions: Is the percentage of African nations among UN members larger or smaller than the number you just wrote? What is your best guess of the percentage of African nations in the UN? Does the random number from the …rst stage in‡uence the answers in the second stage of the experiment?

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

12 / 36

Anchors

The average estimate of those who saw 10 were 25% and that of those who saw 65 was 45%. This phenomenon is called anchoring. It occurs when people consider a particular value for an unknown quantity before estimating that

• quantity. The estimates on average stay close the number people

considered. Anchoring is one of the most reliable and robust results of experimental psychology.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

13 / 36

Anchors

There are two psychological processes that lead to anchoring. The …rst is insu¢cient adjustment. Start from an anchoring number, assess whether it is too high or too low, and gradually adjust your estimate by mentally “moving” from the anchor. Insu¢cient adjustment operates at questions like the following one: What is the boiling temperature of water at the top of Mount Everest? An anchor comes immediately to your mind (100C) from which you adjust downwards (because you know that the boiling temperature must be lower). On average, the adjustment process stops too early.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

14 / 36

Anchors

The second process is priming. Priming is the temporary activation of an individual’s mental representations (ideas) and the e¤ect of this activation on behavior in an unrelated subsequent task. As an illustration, consider the following question. Was Gandhi more or less than 144 years old when he died? How old was Gandhi when he died? It does not make sense to adjust the age downwards from 144. However, it creates the image of a very old person. On the contrary, if 144 is substituted by a very low number, say 12, this creates the image of a very young person.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

15 / 36

Anchors

Unlike many other biases, anchoring can be measured. Consider the following question. Is the height of the tallest redwood more or less than X feet? What is your best guess about the height (in feet) of the tallest redwood? Let XH = 1200 be the “high anchor” and XL = 180 the “low anchor”. The high anchor produces an average estimate of YH = 844, while the low anchor produces an average estimate of YL = 282. The anchoring index is de…ned by anchoring_index = YH YL XH XL %. In the current example, the anchoring index equals 55%, a typical number for many anchoring results.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

16 / 36

Anchors

Anchoring can be strong in the real world. It seems especially relevant to bargaining. Imagine the purchase of a real estate. The …rst value of the bidding process, the seller’s asking

• r listing price, might serve as an anchor, e¤ectively determining the

neighborhood of appropriate prices for subsequent negotiations. Northcraft and Neale (1987) tested this hypothesis both for novice (business students) and expert (professional real estate agents with substantial business experience) participants.6

6Northcraft, Gregory, and Margaret A. Neale (1987): “Experts, Amateurs, and Real

Estate: An Anchoring-and-Adjustment Perspective on Property Pricing Decisions,” Organizational Behavior and Human Decision Processes 39(1), 84 - 97.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

17 / 36

Anchors

Subjects had to assess the value of a house that was actually on the market. They visited the house and received a 10-page packet of information that included an asking price (i.e., the anchor). The low anchor was 74.900 USD, while the high anchor was 83.900 USD. Each subject gave her opinion about a reasonable buying price for the house. They were also asked about the factors that had a¤ected their judgment.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

18 / 36

Anchors

Both types of subjects exhibited strong anchoring e¤ects. The anchoring index was 41% for the experts, and 48% for the students. While experts indicated that the asking price was no source of in‡uence (which obviously was wrong), the students admitted that they were in‡uenced by it. Hence, anchoring can play an important role in situations where people have expertise and where a lot of money is at stake.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

19 / 36

Anchors

Another (frightening) example for anchoring in the real world comes from jurisdiction. German judges with an average of more than 15 years of experience …rst read a description of a woman who had been caught shoplifting and then rolled a pair of dice that resulted in either a 3 or a 9. Then the judges were asked whether they would sentence the woman to a term in prison greater or lesser, in months, than the number showing on the dice. They also had so specify the exact prison sentence they would give to the woman. The anchoring e¤ect was 50%.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

20 / 36

Availability

Suppose you are asked to estimate the frequency of a category such as “the share of people who divorce after the age of 60”. Most likely, you may search for examples for this category. If you can recall many (few) examples, you will estimate the size of the category rather high (low). The availability heuristic is de…ned as the process of judging frequency by “the ease with which instances come to mind”.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

21 / 36

Availability

The availability heuristic substitutes a di¢cult question for a simple

• ne: You wish to estimate the size of a category, but you report an

impression of the ease with which instances come to mind. Clearly, this creates a bias. Here is a list of factors (other than frequency) that make it easy to come up with instances: Salient or dramatic events (terroristic attacks, sex scandals among politicians), personal experiences, pictures, and vivid examples (which are more available than statistics). Resisting the in‡uence of availability biases is di¢cult.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

22 / 36

Availability

Availability matters in marriages (and probably many other joint projects). Ross and Sicoly (1979) asked spouses “How large was your personal contribution to keeping the place tidy, in percentages?” (and similar questions).7 The self-assessed contributions added up to more than 100%. Both spouses remember their own individual e¤orts and contributions much more clearly than those of the other. This leads to a di¤erence in judged frequency. Spouses also overestimated their contribution to causing quarrels (hence, the bias is not self-serving). The study shows that there are egocentric biases in availability. Awareness of these biases may increase the quality of our relationships.

7Michael, Ross, and Fiore Sicoly (1979): “Egocentric biases in availability and

attribution,” Journal of Personality and Social Psychology 37(3), 322 - 336.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

23 / 36

Availability

The following experiment by Schwarz et al. (1991) shows that the ease of retrieval is treated as information.8 Subjects were given the following two tasks. First, list X instances in which you behaved assertively. Next, evaluate how assertive you are. In one treatment X = 6, in the other X = 12. While the …rst instances probably come easily to mind, most people struggle to come up with the last ones to complete the twelve: ‡uency will be low. Obviously, the experience of low ‡uency is stronger for those in the condition with X = 12. What will count more: the amount retrieved or the ‡uency of retrieval?

8Schwarz, Norbert, Herbert Bless, Fritz Strack, Gisela Klumpp, Helga

Rittenauer-Schatka, and Annette Simons (1991): “Ease of retrieval as information: Another look at the availability heuristic,” Journal of Personality and Social Psychology 61(2), 195 - 202.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

24 / 36

Availability

Subjects in the X = 12 treatment on average rated themselves as less assertive than people in the X = 6 treatment. This e¤ect can be turned around. If subjects are asked to list (six or twelve) instances in which they had not behaved assertively, then those in the X = 12 treatment on average rate themselves as more assertive than people in the X = 6 treatment. The experience of diminishing ‡uency is stronger than the pure number of retrieved instances. There are many versions of this experiment that produce entertaining results (people believe that they use their bicycles less often after recalling many rather than few instances; they are less con…dent in a choice when they are asked to produce more arguments to support it, etc.).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

25 / 36

Availability

Unlike anchoring, the e¢cacy of the availability heuristic can be manipulated. For example, people who are personally involved in the judgment are more likely to consider the number of instances they retrieve instead

• f ‡uency.

They conducted a study where half of the subjects had a family history of cardiac disease and were expected to take the task more seriously than the others. The subjects’ task was to recall either three or eight behaviors in their routine that could a¤ect cardiac health. Next, they had to assess their risk of disease.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

26 / 36

Availability

Students with no family history of heart disease were casual about the task and followed the availability heuristic. The students with a family history of heart disease felt safer when they retrieved many instances of safe behavior (or greater danger when they retrieved many instances of risky behavior). In a state of lower vigilance, the availability heuristic is more active (e.g., if subjects are engaged in another e¤ortful task at the same time; if they are in a good mood because they just thought of a happy episode in life; if they score low on a depression scale; if they are novices on the topic of the task, in contrast to true experts; etc.).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

27 / 36

Availability and Risk

The availability heuristic matters for the perception of risk. After a disaster, people are very concerned and are ready to adopt substantial protective measures (insurance purchase, protection actions against ‡oods, etc.). However, memories of disaster become less available over time and so do worries and concerns. In a famous study by Paul Slovic, subjects are asked to consider pairs

• f causes of death: diabetes and asthma, stroke and accidents, asf.

For each pair, subjects indicated the more frequent cause and estimated the ratio of the two frequencies.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

28 / 36

Availability and Risk

Here are some results from this study. Strokes cause almost twice as many deaths as all accidents combined, but 80% of subjects judged accidental death to be more likely. Tornadoes were seen as more frequent killers than asthma, although the latter cause 20 times more deaths. Death by disease is 18 times as likely as accidental death, but the two were judged about equally likely. “The world in our heads is not a precise replica of reality; our expectations about the frequency of events are distorted by the prevalence and emotional intensity of the messages to which we are exposed.”

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

29 / 36

Availability and Risk

Biased reactions to risks may be an important source of erratic and misplaced priorities in public policy. For example, the main channel through which terrorism works is by creating availability. The number of casualties from terror attacks is very small (even for countries with a lot of terrorism such as Israel) when compared to

• ther sources of death.

However, the images or terrorism massively increase availability, creating pressure on policy to act. Accordingly, the investments of many countries on counter-terrorism are out of scale (especially in the United States).

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

30 / 36

Representativeness

Consider the following task. Tom W is a graduate student at the main university in your state. Please, rank the following …elds of specialization in order of the likelihood that Tom W is now a student in each of these …elds. Use 1 for the most likely, 9 for the least likely. business administration, computer science, engineering, humanities and education, law, medicine, library science, physical and life sciences, social science and social work Obviously, the relative size of enrollment (the base rate) is the key to a solution.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

31 / 36

Representativeness

Now, consider the following personality sketch of Tom W written by a psychologist, on the basis of a psychological tests of uncertain quality. Tom W is of high intelligence, although lacking in true creativity. He has a need for order and clarity, and for neat and tidy systems in which every detail …nds its appropriate place. His writing is rather dull and mechanical, occasionally enlivened by somewhat corny puns and ‡ashes of imagination of the sci-… type. He has a strong drive for

• competence. He seems to have little feel and little sympathy for other

people, and does not enjoy interacting with others. Self-centered, he nonetheless has a deep moral sense. Rank the nine …elds of specialization by how similar the description of Tom W is to the typical graduate student in each of the …elds. Use 1 for the most likely and 9 for the least likely.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

32 / 36

Representativeness

This task asks about representativeness. Base rates do not matter. Most people think that Tom W is quite representative for a computer scientist. The average ranking is as follows: computer science, engineering, business administration, physical and life sciences, library science, law, medicine, humanities and education, social science and social work Note that for the purpose of this task, the accuracy of the description (whether it is true or not) is irrelevant.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

33 / 36

Representativeness

The third task is now to rank the …elds of specialization in order of the probability that Tom W is a graduate student in each of the …elds. The subjects of this experiment were graduate students of psychology and knew the relevant statistical facts (i.e., the base rates). They also were aware of the fact that the description is not trustworthy. Would they take the base rates into account or only use representativeness?

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

34 / 36

Representativeness

The average ranking of the …elds did not di¤er from the rating by similarity to the stereotype. Hence, subjects only used representativeness in their assessment. This e¤ect is called substitution: The (di¢cult) probability question is substituted by the (simple) representativeness question. Obviously, ignoring the base rates is a mistake.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

35 / 36

Representativeness

One consequence of judgments by representativeness is an excessive willingness to predict the occurrence of unlikely (low base rate) events. Another consequence is an insensitivity to the quality of evidence. We automatically process the information available as if it were true. Kahneman suggests two keys to discipline intuition: (1) Anchor your judgment of the probability of an outcome on a plausible base rate. (2) Question the diagnosticity of your evidence.

• Prof. Dr. Heiner Schumacher (KU Leuven)

Incentives and Behavior

• 9. Behavioral Biases I

36 / 36