Anchoring Effect: How Gullible Are We?

Transcription

Anchoring Effect: How Gullible Are We?
Introduction
Anchoring Effect: How
Gullible Are We?
My project is about something called the
anchoring effect. The anchoring effect is a
psychological phenomenon that is used in
everyday life, from guessing rice in a jar to
buying and selling a house. It is very
important to understand how anchoring
affects people to improve the every day's
decision making.
By Matias Badino
Photographs
Hypothesis 1: there is a positive correlation
between the anchor and the answer
provided.
(A) Jar with rice
(B) Clipboard with collected data
People are asked to estimate the number of grains in a jar (see Picture A). The question
asked is:
Do you think there are more or less than X grains in this jar?
How many grains do you think there are?
●
Very large (up to 20 million) and low estimates
(as low as 371) were given for all anchors. I
have decided to use the median instead of the
mean to summarize the data to avoid the effect
of those outliers.
As a criteria to evaluate the bias, I calculate the
ratio of the biased answer with the answer of
the control question rounded to the nearest
integer. This number is called bias factor. A bias
factor of 10 means that the biased answer is 10
times larger than the one given without anchor.
where X is the anchor. Anchors for 1K, 10K, 100K, and 1M are used. A control question is
also asked without providing any anchor (without the first question).
Conclusions
People of different ages and gender are surveyed at different days, times of the day, and
places.
In this experiment, the anchoring bias was
huge. In general, interviewees were highly
biased when the question was loaded with an
anchor. The bias factor is small (between 1 and
4) when the anchor is close to the control, but it
can be up to more than 40 when the anchor is
1 million. These results confirm my first
hypothesis.
Data
Anchor/Gender All
Males
Females
Control
5,250
6,500
3,513
1,000
3,000
4,000
2,750
10,000
11,300
9,000
12,300
100,000
55,000
36,664
69,101
1,000,000
200,250 128,500 375,000
Rationale H1: estimating the number of
grains in a jar is very difficult. Therefore, I
expect the interviewee to be strongly
influenced by the anchor provided.
Anchor/Gender All
Males
1,000
1
10,000
2
100,000
10
1,000,000
38
Rationale H2: there is no conclusive
evidence that demonstrate superiority in
numerical estimation between genders [2],
therefore, I expect the same anchoring
effect in males and females.
Jars (3), rice (2 pounds), scale (1),
computer (1), clipboards (3), paper,
printer (1), scissors (1), booklet (1), glue,
reading material [1], spreadsheet tool,
Internet.
(C) Carnegie Science Center
Procedure
Hypothesis 2: Males and females are
equally affected by the anchor provided.
1
1
6
20
Females
1
4
20
107
120
100
All
Males
Females
80
Bias Factor
Materials
●
286 people were surveyed over the period of
16 days.
133 interviewees were males and 153 were
females
Age range was 6 to 75 years old.
Data was collected at the Carnegie Library,
Carnegie Science Center, Pittsburgh Gifted
Center, Pittsburgh Minadeo, and at the
streets of Squirrel Hill.
Evaluation Criteria
Problem
Hypotheses
●
●
Anchoring effect is the bias in the decision
of a person by the previous introduction of
an anchor. The anchoring effect has been
proven
on
several
psychological
experiments [1]. In this project, I provide
further proof of the anchoring effect and
evaluate whether it affects differently males
and females.
Determine if there is a anchoring effect
when people is asked to estimate the
number of grains in a jar full of rice (see
Section Procedure). Further determine if
there is a difference of the bias between
males and females.
Data Collection Efforts
60
40
20
0
1,000
10,000
100,000
Anchor [number of grains]
1,000,000
The table shows the median answer
provided by the interviewees for the
control questions and each one of the
anchors
Results
The table and graph show the bias factor for
each of the anchors, for males, females, and
both together. There is a positive strong
correlation of the bias factor with the anchor.
This means that the larger the anchor, the
larger the bias.
Observe that the anchoring effect is very large
in this experiment. Females provided an
answer that was up to 107 times larger than
the answer they provide when no anchor is
given.
Females were far more influenced by the
anchoring effect than males. With the exception
of the anchor 1,000, females bias factor was 4
times larger than those of males. This rejects
my second hypothesis. This result seems to be
in accordance with other studies [3].
Future Work
More detailed analysis will be performed using
the recorded age of the interviewees and the
recorded time taken to provide the answers.
References
[1] D. Kahneman, “Thinking, Fast and Slow”, Publisher: Farrar, Straus and
Giroux, 2013.
[2]
Sex
differences
in
human
psychology,
Wikipedia
at
http://http://en.wikipedia.org/wiki/Sex_differences_in_human_psychology.
[3] A. Kudryavtsev and G. Cohen, “Behavioral Biases in Economic and
Financial Knowledge: Are They the Same for Men and Women?”, in
Advances in Management & Applied Economics, vol.1, no.1, 2011.