Heuristics and Biases in Decision Making

Tversky and Kahneman (1974)

Basic decision making

• Many decisions are based on beliefs concerning the likelihood of uncertain events
such as the outcome of an election, the guilt of a defendant, or the future value
of a dollar.
• Beliefs are expressed in terms of subjective probabilities.
• Subjective assessment of probabilities is similar to subjective assessment of other
physical entities such as distance, size, etc.
• Distance – Objects that appear sharpers are assumed to closer to you than the
objects that appear blurred. And this is a good heuristic.
• Except on days when the visibility is poor, or even on days when it is a bright and
sunny day!!!

2
Who is Steve?

“Steve is very shy and withdrawn, invariably helpful but with little interest in
people, or in the world of reality. A meek and tidy soul, he has a need for order and
structure, and a passion for detail.”

What is a probability that Steve is a farmer, a salesman, a pilot, a librarian or a

physician?

Steve is most likely to be…

3
Now, Who is Dick?

“Dick is a 30 year old man. He is married with no children. A man of high ability and
high motivation, he promises to be quite successful in his field. He is well liked by
his colleagues.”

Is Dick an engineer or a lawyer?

What is the probability that Dick is an engineer?

4
Now, Who is Dick?

• It is known that Dick was selected from a sample of 100 professional that had 70
engineers and 30 lawyers. Does this information alter your earlier answer?

• Should it?

5
Hospital variability

A certain small town is served by two hospitals. In the larger hospital about 45 babies are
born each day, and in the smaller hospital about 15 babies are born each day. As you know,
about 50% of all babies are boys! However, the exact percentage varies from day to day.
Sometimes it is higher than 50%, sometimes lower.
For a period of 1 year, each hospital recorded the days on which more than 60% of the
babies born were boys. Which hospital do you think recorded more such days?

The larger hospital

The smaller hospital
About the same (i.e. within 5% of each other)
6
Churning the urn!!

Imagine an urn filled with balls, of which 2/3rd are of one color, and 1/3rd of another.
One individual has drawn 5 balls from the urn, and found that 4 were red and 1 was
white. Another individual has drawn 20 balls and found that 12 were red and 8
were white.

Which of the two individuals should feel more confident that the urn contains 2/3rd
red balls and 1/3rd white balls, rather than the opposite?

7
Representativeness

8
Representativeness heuristic

• Many times we face the question of what is the probability that object A belongs
to class B? What is the probability that event A belongs to process B?
• In these situations we typically rely on representativeness heuristics.
• We evaluate probabilities by the degree to which A is representative of B, that is
how closely A resembles B!!
• If A resembles B closely, the probability that A originates from B is estimated to
be high.
• This can lead to serious errors in judgement. For, representativeness doesn’t
consider several factors that should affect judgement of probability.

9
Insensitivity to priors

• Estimating the probability that event A belongs to process B should be guided by

the prior probability (or prior belief).
• Representativeness heuristics completely ignores the prior probability.
• Recall the librarian example (Steve) and the Engineer\Doctor example.
• Most of the time, we rely on the similarity of the “object” with the “class”
ignoring the proportions.
• The fact that there are more farmers than librarians in the population should
enter into any estimate of probability that Steve is a librarian.
• Similarly, with the engineer\doctor example!!

10
Average height estimation

• Suppose that average height of Indian female is 5 ft. (Study by Imperial College
London, 2016)
• What is the likelihood that the average height in a random sample of 10 women
will be more than 5 ft?
• What is the likelihood that the average height in a random sample of 1000
women will be more than 5 ft?
• What is the likelihood that the average height in a random sample of 100 women
will be more than 5 ft?

11
Insensitivity to sample size

•  Estimation of the population mean from the sample mean should be guided by
the sample size. (Remember the inferential statistics principle that things are
good as long as “”.)
• Studies have shown that probability of obtaining an average height more than 5 ft
was assigned same values for samples of 1000, 100 or 10 women.
• Recall also, the hospital child birth problem stated earlier.
• Subjects typically judge the probability of obtaining more than 60% boys to the
same for small and big hospital, probably because these events are described by
the sample statistic and therefore equally representative of the general
population.

12
Insensitivity to sample size

• Recall the urn problem.

• The correct posterior odds are 8:1 for the 4:1 sample and 16:1 for 12:8 sample.
• However, many feel that first sample provides a stronger evidence of the
hypothesis that the urn is predominantly red, because the proportion of red in
the first sample is more than the second sample.
• Intuitive judgement seems to be guided more by the sample proportion, and
seems to be unaffected by sample size!

13
While gambling….

• Here is a sequence of Heads and Tails… T-H-T-H-T-T-H. What is the probability that
the next toss would result in a “T”?

• Take this sequence now… H-H-H-H-H-H. What is the probability that the next toss
would result in a “T”?

14
Misconception of chance

• People expect that a sequence of events generated by a random process will

represent the essential characteristics of that process even when the sequence is
short.
• People think H-T-H-T-T-H is somehow more likely than H-H-H-T-T-T.
• Second sequence doesn’t appear random first of all! Secondly, it seems more
likely than the sequence H-H-H-T-H which doesn’t represent the fairness of the
coin!
• Chance is viewed as a self-correcting process in which a deviation in on direction
induces a deviation in the opposite direction to restore equilibrium.

15
Description of two companies

• Company A – IIT graduate CEO, new technology, smart workforce, etc.

• Company B – old family run business, hardly any technology, dirty/greasy
workers.

Which company is more profitable?

16
Insensitivity to predictability

• If the description of the company is favourable, a very high profit will appear
most representative of that description, if the description is mediocre, a
mediocre performance will appear more representative.
• The people predict solely based on the favourableness of the description, their
prediction will be insensitive to the reliability of evidence and to the expected
accuracy of the prediction.
• This also violates the normative statistical theory.
• When predictability is nil, same prediction should be made in all cases.
• Example: Predicting success of a teacher 5 years from today, using a lecture
today.

17
Illusion of validity

• People express great confidence in the prediction that a person is librarian when
given a description of his personality which matches the stereotype of librarians,
even if the description is scanty, unreliable, and outdated.
• The unwarranted confidence which is produced by a good fit between the
predicted outcome and the input information may be called the illusion of
validity.
• People express more confidence in predicting the final grade-point average of
student whose first-year record consists entirely of B’s than in predicting the GPA
of a student whose first-year record includes many A’s and C’s.

18
Flight training…

• Flight instructors note that praise for an exceptionally smooth landing is typically
followed by a poorer landing on the next day.
• While harsh criticism after a rough landing is usually followed by an improvement
on the next try.
• So instructors concluded that verbal rewards are bad and verbal punishments are
beneficial.
• Are they correct?

19
Misconception of regression

• People don’t completely understand the power of regression.

• They don’t expect regression in many contexts where it is bound to occur.
• Secondly, when they recognize the occurrence of regression, they often invent
spurious causal explanation for it.
• Flight instructors observations are unwarranted because of the regression
towards the mean.
• Over repeated examinations, a improvement will usually follow a poor
performance, and a deterioration will follow an outstanding performance.

20
More questions for you to ponder…

•  You are asked for sample words (three letters or more) at random from English
text. Is it more likely that the word starts with r or that r is the third letter?
• You are asked to rate the frequency with which abstract words (thought, love)
and concrete words (door, water) appear in written English. What is your guess?
• Consider a group of 10 people who form committees of k members, . how many
different committees of k members can be formed?
• If k = 2?
• If k = 8?

21
Availability heuristics

22
Availability heuristics

• At times, people assess frequency of a class or the probability of an event by the

ease with which instances or occurrences can be brought to mind.
• e.g. prevalence of heart-attack?
• This judgement heuristics is called availability.

• In theory, availability is a useful clue for assessing frequency or probability,

because instances of large classes are usually recalled better and faster than
instances of less frequent class.
• However, this heuristics leads to some predictable biases…

23
Biases due to retrievability

• When the size of a class is judged by the availability of its instances, a class whose
instances are easily retrieved will appear more numerous than a class of equal
frequency whose instances are less retrievable.
• In a study, participants were told a list of well-known personalities and asked
whether the list contained more men than women.
• Different list were presented to different sets of participants – some lists
contained men who were more famous than the women and in some lists vice
versa.
• In each of the lists, the subjects erroneously judged that the class (gender) that
had the more famous personalities was more numerous!!

24
Biases due to retrievability: Seeing is believing!

• In addition to familiarity, there are other factors like salience which affects
retrievability of instances.
• Impact of seeing a house on fire on the subjective probability of such instances, is
probably greater than the impact of read about fire in the local newspaper.
• Also, recent occurrences are more like to be relatively more available than earlier
occurrences.

25
Biases due to the effectiveness of a search set

• Recall the exercise where you were searching for English words or three letters or
more…
• Because it is easier to search for words by their first letter than by their third
letter, most people judge the words that start with a given consonant to be more
numerous than words in which the same consonant appears in the third position.
• They do so even for consonants such as r or k, that are more frequent in the third
position than in the first.

26
Biases due to the effectiveness of a search set

• Recall the exercise on searching for abstract words and concrete words.
• An easy way to answer that question is to search for contexts in which the word
would appear.
• It seems easier to think of contexts in which an abstract concept is mentioned
than to think of contexts in which a concrete word is mentioned.
• If the frequency is judged by availability of the contexts in which they appear,
abstract words will be judged as relatively more numerous than concrete words.

27
Bias of imaginability

• Sometimes
  we need to assess the frequency of a class whose instances are not
stored in memory but can be generated according to a given rule.
• In such cases, we typically generate several instances and evaluate frequency or
probability by the ease with which the relevant instances can be constructed.
• However, ease of constructing the instances doesn’t always reflect the true
frequency!
• Recall the exercise of creating a committee of 8 people from a group of 10…
• Correct answer is . Therefore, the number of committees of k members equals
the number of committees of (10 – k).

28
Bias of imaginability

• One way to answer the question to mentally construct committees of k members

and to evaluate their numbers by the ease with which they come to mind.
• Committees of fewer members, say 2, are more available than the committees of
many members, say 8.
• We usually partition the set into disjoint sets of required number. Since there are
more disjoint sets of 2 members, we judge the committees of 2 members to be
numerous than the committees of 8 members.

29
Illusory correlation

• Judgement of how frequently two events co-occur could be based on the

strength of bond between them. When the association is strong, one is likely to
conclude that the events have been frequently paired.
• e.g. What is the probability of someone eating samosa and drinking tea together?

30
Auto rickshaw ride…

• You stop an auto rickshaw on the roadside and ask the driver to quote the fare to
go from place A to place B.
• The driver quotes, say INR 150.
• After haggling, the deal is struck at INR 120. And you are happy that you saved
INR 30.

• What if the driver had quoted INR 130? Would you still be happy that the deal
was stuck at INR 120?

31
Adjustment and Anchoring

32
Adjustment and Anchoring

• In many situations, people make estimates by starting from an initial value that is
adjusted to yield the final answer.
• The initial value, or the starting point, may be suggested by the formulation of
the problem, or it may be the result of a partial computation.
• In either cases, adjustments are typically insufficient! That is, different starting
points yield different estimates which is biased towards the initial value.
• We call this anchoring.

• As with all the heuristics, this one also results in various biases…

33
Insufficient adjustment

• Subjects were asked to estimate percentage of African countries in the UN.

• When one group was given a random number of 10, the estimate was 25.
• Other group that received a random number of 65, estimated the percentage to
be 45!
• Even when the subjects were paid for accuracy, the estimates didn’t get any
better.
• The estimates get anchored at the number the group received.

34
Anchoring

• Group of high school students were asked to estimate the following in 5

seconds…
• What is 1x2x3x4x5x6x7x8 ?
• What is 8x7x6x5x4x3x2x1 ?
• To rapidly answer the question, people perform few steps of computation and
estimate the product by extrapolation or adjustment.

• Estimate for the former was 512, the estimate for later was 2,250. Correct answer
is 40,320!!

35
Bias in evaluation of conjunctive and disjunctive events

• Participants were asked their preferences for betting out of the three options:
1. Simple event – drawing a red ball from a bag containing 50% red balls and 50%
white balls.
2. Conjunctive event – drawing a red ball seven times in succession, with
replacement from a bag containing 90% red balls and 10% white balls.
3. Disjunctive event – drawing red ball at least once in seven successive tries, with
replacement from a bag containing 10% red balls and 90% white balls.
• Before we proceed, which bet would you prefer?

36
Bias in evaluation of conjunctive and disjunctive events

• Significant majority of participants preferred conjunctive event (probability of

which is 48%) over a simple event (probability of which is 50%).
• Also, participants preferred simple event (p = 0.5) over disjunctive event (p =
0.52).
• The reason? Anchoring
• Stated probabilities set the anchor (and a good starting point) in calculating the
probabilities for conjunctive and disjunctive event.
• In real life, project planning is a conjunctive event (all the components needs to
succeed for the successful completion). Whereas, an industrial accident is a
disjunctive event (any of the component needs to fail for a disaster)!

37