Why People Make Bad Decisions: The Role of Cognitive Biases

L6σ

Why People Make Bad Decisions
The Role of Cognitive Biases

Scott Leek
Sigma Consulting Resources, LLC

American Society for Quality Denver Section
October 16, 2012

© 2012 Sigma Consulting Resources, LLC 1

Cognitive Biases & Decision Making L6σ
Topics
•  Cognitive Biases and decision making

•  Review of Common Biases and Mitigation Strategies
  Hindsight Bias
  Confirmation Bias
  Anchoring and Adjustment Heuristic
  Availability Heuristic
  Representativeness Heuristic
o  Insensitivity to Sample Size
o  Insensitivity to Prior Probability
o  Conjunction Fallacy

•  Decision Quality Control Checklist


Exercise L6σ
Decision Making

Identify at least 3 decisions you have made, or been
involved in making, that turned out to be wrong or “not so
good.” The decisions can be recent or in the past, large or
small.


Cognitive Bias L6σ
Definition
•  “…a replicable pattern in perceptual distortion,
inaccurate judgment, illogical interpretation, or what is
broadly called irrationality”

•  Arise from multiple confounded sources
  Information-processing shortcuts (e.g., availability heuristic)
  Mental noise (wrong way on a one-way street)
  Limited information processing capacity (e.g., Bayesian
probabilities)
  Emotional or moral motivations (e.g., just-world hypothesis)
  Social influence (e.g., groupthink)


Exercise Test L6σ
Decision Making
By a show of hands how people…

Identified three or more examples of “not so good decisions?”

Identified two or more examples of “not so good decisions?”

Identified one or more examples of “not so good decisions?”

If you were unable to identify an example you may be
suffering from the…


Cognitive Biases L6σ
Hindsight Bias


Hindsight Bias L6σ
Definition

In hindsight…

•  “…the consistent exaggeration of what could have been
anticipated in foresight” (“I knew it all along” or “creeping
determinism”)

•  “…the inclination to see events that have already
occurred as being more predictable than they were
before they took place”


Hindsight Bias L6σ
Problems (“So What”)
•  Failure to learn from the outcome of events, not being surprised by
anomalous outcomes (if we are unable to acknowledge when our
predictions are wrong, they will never be right)

•  Influences attributions of blame after unforeseen catastrophic
events

•  People tend to misremember (memory distortion) their predictions
in order to exaggerate in hindsight what they knew in foresight

•  Causes people to rely too heavily on knowledge of the outcomes of
historical events, leading to accepting sufficient, though not
necessary explanations too easily (“tried it, didn’t work” turns out
there was an interaction)


Hindsight Bias L6σ
Preventing or Mitigating
•  Awareness is not enough to mitigate

•  Use of the scientific method or derivative like Plan-Do-Study-Act
(PDSA)

•  Recording predictions prior to events (a priori) like process
changes, experiments, et cetera and reviewing those predictions
after the events (a posteriori), formally updating current knowledge

•  Focus on why outcomes occur, not just if the predictions are
correct, try to explain alternative or anomalous outcomes

•  Reward people based on logic of judgment, not just outcomes
(Hogarth, e.g., control charts and Type I errors, testing with true/
false give a reason)


Confirmation Bias


Confirmation Bias L6σ
Case Study
A sales manager believes that a new marketing method
will increase the sales call success rate. An experiment
was designed to test the effectiveness of the new method.
The experiment was run for one week when 480 sales
calls were made. The new method was randomly assigned
to sales calls and the number of sales made was recorded.
The brochure resulted in 270 sales.

Treatment # Sale Made # No Sale Made
New Method 270 90
Old Method 90 30

Conclusions?

Definition
•  A tendency for people to favor information that confirms existing
beliefs or theories (paradigm, Kuhn)

•  Ambiguous evidence is interpreted as supporting existing beliefs or
theories

•  Fail to search for disconfirming evidence

•  Typically falls into three categories of bias:
  Search for information
  Interpretation
  Memory (hindsight bias)

•  In light of the confirmation bias the oft quoted “I’ll believe it when I see it”
might better be stated “I’ll see it when I believe it.” (see Thomas Kuhn, The
Structure of Scientific Revolutions


•  Overconfidence in decision-making based on ignoring or not
seeking all relevant data

•  Leads to flawed causal models which in turn influences what we
observe, leading to flawed causal models in what can be a self-
reinforcing loop (Senge’s “Reflexive Loop”)

•  Leads to the “We have made the decision now find the data to
support it…” scenario

•  Plays a role in “Groupthink”


Problems (“So What”) I take actions
based on my
beliefs
I adopt beliefs
about the world
I draw
conclusions
The “Reflexive Loop” I make
(our beliefs affect what assumptions
data we select next time) based on the
meanings I added
I add
meanings

I select data from
what I observe
Observable data
and experiences
From the “Fifth Discipline Field Book” by Peter Senge

•  Beware of asking (or being asked) to “prove” something. When the
objective is to “prove,” that will be the bias

•  Build into questions, data collection and analysis a search for
disconfirming information (use all quadrants of the 2X2 table)

•  Adopt the opposing or contrary point of view or position, in a group
allow someone to play “devils advocate”

•  Use of the scientific method or derivative like Plan-Do-Study-Act
(PDSA)


Anchoring and Adjustment


Anchoring and Adjustment L6σ
Case Study
An engineer was asked to prepare a budget for completing
engineering projects over the next year. The engineer
obtained the budget for the previous year and after a brief
analysis prepared a budget similar to the previous year’s
budget but 5% higher.

What was the basis for the budget (goal)?


Definition
•  In the process of making estimations…”people start with an
implicitly suggested reference point (anchor) and make adjustments
to it to reach their estimate,” even if the anchor is irrelevant

•  In an early study (Tversky and Kahneman) spun a roulette wheel in
front of a group of experimental subjects. The result was 65.
Subjects were asked to record this result. They were then asked to
estimate the percentage of African nations that were members of
the United Nations. The process was repeated with a second group
of subjects, but the result from the roulette wheel was 10

•  The median estimates for the two groups were significantly different
with the group shown the 65 having a median estimate of 45% and
the group shown the 10 having a median estimate of 25%


•  May lead to frustration and failure to accomplish goals and
objectives because the goal was not realistic or attainable

•  May lead to the waste of underachievement, much more could have
been accomplished if the goal was set higher

•  Application and implications for process improvement teams or
functional teams with measureable goals


•  Anchors analogous to “last years budget” will always influence
estimates but can be balanced by an exploration of the causal
factors influencing the estimate (outcomes)

•  Use of models like the SMART (Specific, Measureable, Attainable,
Relevant, Time-bound) criteria when creating goals

•  Can have profound implications when negotiating


Availability Heuristic


Availability Heuristic L6σ
Case Study
Structure A Structure B

XXXXXXXX XX
XXXXXXXX XX
XXXXXXXX XX
XX
A path in a structure is a line that XX
connects an element in the top XX
row to an element in the bottom XX
row and passes through one and
XX
only one element in each row.
XX
In which structure (A or B) are The number of paths in each
there more paths? How many? structure is the same 83 = 29 = 512


Case Study
•  In a study (Tversky and Kahneman) 85% of respondents found
more paths in Structure A than in Structure B

•  The bias towards Structure A is explained by the eight columns
which make the paths more distinctive and available


Definition
•  “[A] mental shortcut that uses the ease with which examples come
to mind to make judgments about the probability of events. The
availability heuristic operates on the notion that ‘if you can think of
it, it must be important’”

•  How many words start with the letter “k?” How many words have
the third letter of “k?”

•  The heuristic can be beneficial, but the frequency that events come
to mind are usually not accurate reflections of their actual
probability in reality


•  If the available instances or associations reasonably represent the
circumstances there is not a problem, otherwise correct conclusions
and decisions are more a matter of good fortune

•  Customers and stakeholders are often surveyed about their
experience's and perceptions regarding a product or service, the
responses can often be biased by the availability heuristic


•  Prior to decisions check the data used to make the decision, was
the most available data used? If so, is there bias?

•  Is the data used to make the decision representative?

•  Is base rate data available?


Representativeness Heuristic


Representativeness Heuristic L6σ
Definition
•  Used when making judgments about the probability of events often
because of its ease of computation

•  Representativeness is "the degree to which [an event] (i) is similar
in essential characteristics to its parent population, and (ii) reflects
the salient features of the process by which it is generated”


Representativeness Heuristic L6σ
•  Just because something is more representative does
not mean it is more likely (base rate vs. case rate data)

•  People overestimate their ability to predict the likelihood
of an event

•  Rooted in three types of biases
  Insensitivity to sample size
  Insensitivity to prior probabilities
  Conjunction fallacy


Insensitivity to Sample Size L6σ
Case Study
A certain 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
girls. However, the exact percentage varies from day to day.
Sometimes it may be 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 girls. Which hospital do you think
recorded more such days?

A. The larger hospital
B. The smaller hospital
C. About the same (that is, within 5% of each other)


Case Study
•  In a study (Tversky and Kahneman) 56% of respondents select
option C, and 22% selected options A and B respectively

•  According to sampling theory the larger hospital is much more likely
to report a ratio close to 50% on a given day compared to the
smaller hospital


Definition
•  Tendency to expect different sized groups of samples to be equally
representative of a process or population

•  Insensitivity to, or lack of knowledge of the role sampling error plays
in sample statistics


Insensitivity to Prior Probability L6σ
Case Study
•  A study (Tversky and Kahneman) involved telling one group of
participants that a provided description of a person came from a
group of 70 engineers and 30 lawyers and then asking them to
assess the probability that the described person was an engineer
(or lawyer).

•  A second group was told that the description came from a group of
30 engineers and 70 lawyers and asked to assess the same
probability.

•  The experiment was repeated with variations in descriptions.


Case Study
•  Tversky and Kahneman found a strong tendency for participants to
disregard the “base rate” (frequency of occurrence) information
preferring to rely on the descriptive information

•  When considered the base rate probabilities were not adjusted
appropriately (Bayesian probabilities) given the additional
information


Definition
•  Tendency to ignore or improperly weight base rate probabilities

•  Improperly weighting additional information when discounting base
rate probabilities

•  A related bias is the Conjunction Fallacy which states that the
conjunction of two events cannot be more likely than the
occurrence of either event alone (“Linda” study)


Now What?


Now What?
Kahneman, Lovallo, and Sibony have proposed a Decision Quality
Control Checklist involving three phases of assessment

Preliminary Challenge Evaluation


Decision Quality Control Checklist L6σ
Preliminary
1.  Check for Self-interested Biases (overoptimistic)

2.  Check for the Affect Heuristic (in love with the solution)

3.  Check for Groupthink (dissenting opinions explored)


Challenge
4.  Check for Representativeness Bias*

5.  Check for the Confirmation Bias (credible alternatives)

6.  Check for Availability Bias (imagine perfect information)

7.  Check for Anchoring Bias (where did the numbers come from)

8.  Check for Halo Effect (assumption success will be transferable)

9.  Check for Sunk-Cost Fallacy (overly attached to history)

*Kahneman et al refer to this as the Saliency Bias

Evaluation
10.  Check for Optimistic Biases (game it)

11.  Check for Disaster Neglect (worst case bad enough)

12.  Check for Loss Aversion (overly cautious)


Summary
•  Cognitive Biases and decision making

•  Review of Common Biases and Mitigation Strategies
  Hindsight Bias
  Confirmation Bias
  Anchoring and Adjustment Heuristic
  Availability Heuristic
  Representativeness Heuristic
o  Insensitivity to Sample Size
o  Insensitivity to Prior Probability
o  Conjunction Fallacy

•  Decision Quality Control Checklist


Questions


Design of Experiments (DOE) L6σ
References
Kahneman, D., Lovallo, D., Sibony, O., “The Big Idea: Before You Make That Big
Decision…”, Harvard Business Review June 2011, Harvard Business Publishing.

Lovitt, M. R., “Pragmatic Knowledge and Its Application to Quality,” 1992 ASQC Quality
Congress Transactions, ASQ (formerly ASQC), Milwaukee, WI 1992.

Lovitt, M. R., “Using Quality Tools and Methods to Reduce Bias in Judgment,” Quality
Engineering 8(1), 93-116 (1995-96), Marcel Dekker, Inc. 1995.

Moen, Ronald D., Nolan, Thomas W., Provost, Lloyd P., (1991): Improving Quality Through
Planned Experimentation, McGraw-Hill, New York.

Other References
Bazerman, M. A., Judgment in Managerial Decision Making, John Wiley and Sons, New
York, 1990.

Hograth, R., Judgment and Choice, John Wiley and Sons, New York, 1987.

Kahneman, D., Slovic, P., and Tversky, A., Judgment Under Uncertainty: Heuristics and
Biases, Cambridge University Press, Cambridge, 1982.

Why People Make Bad Decisions: The Role of Cognitive Biases

More Related Content

What's hot

Similar to Why People Make Bad Decisions: The Role of Cognitive Biases

Recently uploaded

Why People Make Bad Decisions: The Role of Cognitive Biases