Uploaded by Dhanakshitha KODITUWAKKU

Dhana-psych IA-Final3

advertisement
The effect of the anchoring bias on the
decision making of humans
IB Psychology:Standard Level Internal Assessment
Candidate: *your six-digit alphanumeric personal code e.g. xyz123*
Group: *the personal codes of your group members*
Session: May 2022
Date Submitted: 01.06.21
Word count: 2329
Table of contents
I.
Introduction
3
II.
Exploration
5
III.
Analysis
7
IV.
Evaluation
9
V.
Work cited
11
Appendix i-form of consent
12
Appendix ii-Materials(questions)
13
Appendix iii-Debrief
14
Appendix iv-Raw data
15
Appendix v-Mann-Whitney test results
16
VI.
VII.
VIII.
IX.
X.
2
Introduction
In this technologically developed era, we are subjected to countless sources of information
from which we must draw conclusions. This high cognitive load requires our brains to
function more efficiently to enable the average human to perform their daily routine
effectively. One such way the human brain attempts to increase the efficiency of its
functioning is through its unique decision-making process which has been adapted to make
decisions and choices based on minimal effort. For example, if you were asked “ what’s 2+3”
you would answer with “ 5” without any critical thinking. However, if you were asked “
what’s 19 x 27” you would have to carefully calculate the answer in your head to provide an
accurate answer.
This efficient thinking mechanism was termed system 1 thinking according to the
dual-process model which was developed by Stanovich and West(2000). According to this
model of decision making and thinking humans have two distinct modes of thinking termed
intuitive thinking(system1 ) and rational thinking (system2). Moreover, Stanovich and West
proposed that system 2 is based on careful thought of all possible alternatives for a given
problem while system 1 is based on heuristics.
A heuristic is an assumption or a simple rule which we use to generate quick responses on
changes in our environment and their causations, and form prognoses. We make use of these
‘mental shortcuts’ inorder to make decisions and choices efficiently and effortlessly(cherry,
2021). However, this could also lead to consistent yet somewhat inaccurate patterns of
thinking, termed as cognitive biases. One such bias is the anchoring bias, which is a"biased
estimate toward an arbitrary value considered by judges before making a numerical estimate"
( Strack and Mussweiler,1997 p 437). For example, a man buys a shirt for 50$. Later he
decides to resell his shirt to a customer for 100$. The customer will begin bargaining as he
believes 100$ is overpriced, yet he will unintentionally be anchored to the 100$ value. After
bargaining with the customer for a while the man decides to sell it for 80$. The customer will
be pleased as he will believe that it was an amazing deal as he can pay 20$ less, when in
reality he is paying 30$ more. Nevertheless, anchoring bias is essential in the efficiency of our
decision making processes as it allows us to make reasonable estimates based on limited
information.
The anchoring bias was first experimented with by Strack and Mussweiler(1997). The study
used 69 German university students who were asked to sit at one of four computer
terminals and answer a series of questions. The question was "Did Mahatma Gandhi die
before or after the age of X?" The second question was "How old was he when he died?" The
suggested age for each terminal was different. The anchor values were implausible low
anchor (9); implausible high anchor (140); a plausible low anchor (64);
plausible
high
anchor (79).
3
The results showed that the plausible anchors had a visible effect on participants' estimates,
with a mean of 99.6 years for the plausible high anchor and a mean of 67.9 for the plausible
low anchor. The implausible high anchor had a lesser effect than the implausible low anchor,
with a mean of 66.7 and 50.1 respectively.
We decided to replicate this study, however, changing the famous person to Chadwick
Boseman, A famous actor whom we assumed to be well known among high school students,
and observing the effect of only a plausible high anchor for simplification. Additionally, a
control condition was created to observe the thinking and decision making pattern of
participants without an anchor since the original study did not attempt to observe this.
We aimed to observe the effect a high plausible anchor has on the estimation of the age of
death of Chadwick Boseman among high school students. This is because it would be of high
interest to identify if the anchoring bias affects people who already have a general
understanding or knowledge regarding a specific situation, incident or event. Although the
original study used a famous person (Mahatma Gandhi) as a part of their experimentation it
could be inferred that at the time the experiment was carried out there would have been a slim
chance of the German participants to know about the existence or a general idea of such a
person.
Null hypothesis: The difference between the estimated age of death given by the participants
subjected to the high plausible anchor and the participants in the control condition will be
insignificant.
Research hypothesis: The age estimated by the participants subjected to the high plausible
anchor will be higher than that of the control group.
The independent variable(IV) will be the value assigned as the high plausible anchor while
the dependent variable (DV) is the estimated age of death of Chadwick Boseman.
4
Exploration
This experiment used an independent measures design, which is a design technique where in
each condition of the experiment, different people are employed. Additionally random
allocation is utilized to allocate a condition for each participant(Independent Groups |
Psychology, 2018), since it ensures that each participant has an equal probability of being
assigned to one of two groups. The reason this technique was used is that asking the control
question and the anchor question from the same participant would be redundant as they
would give the same answer for both questions thus making it impossible to observe the
effect of the anchor on their thinking and decision making. Hence we randomly allocated the
participants to a control condition (participants received the control question) and an anchor
condition (participants received the question with the high plausible anchor) by shuffling the
question set before distribution and handing each participant the question sheet on the top of
the set to minimize the effects of individual differences of participants on the estimation of
the age of death of Chadwick Boseman.
The participants for the experiment were acquired through opportunity sampling, which is a
sampling technique where members of a population of interest are recruited by asking if they
would like to take part in research(Mcleod, 2019). Using this technique we recruited 28 high
school students of ages 17 to 19 who did not take psychology as a subject from a library at an
international school in Singapore. The reason behind this selection was because the original
study recruited university students, which suggests that the participants were directly involved
in higher education and were of early age and thus it would be beneficial to use similar
participants in the recreation of the original study. Moreover, recruiting participants who did
not take psychology reduced the chance of demand characteristics which could reduce the
reliability of the results. Finally, We made sure to ask all students if they would like to
participate starting from the entrance of the library until the required limit of participants
were met to reduce selection bias, which is a kind of error commonly seen during opportunity
sampling, when the researcher selects the participants going to be studied(Selection Bias,
2014).
The most significant variable subjected to control was the time the participants were allowed
to answer the question. The participants were given only 20 seconds to read and answer the
question. This was done to confirm that the participant will use their system 1 thinking rather
than their system 2 thinking. Additionally, Chadwick Boseman was used as the famous
person throughout the experiment to minimize the effect of individual differences which
could occur as a result of using multiple famous people throughout the experiment.
The nature of the question provided to each condition of this experiment was similar to the
original study except the famous person used to strengthen the extent to which the original
study was replicated(Appendix ii). The question received by the anchor condition participants
5
was, “ Did Chadwick Boseman die before or after the age of 90”, while the question received
by the control condition participants received, “At what age did Chadwick Boseman die?”
The following steps were taken to perform the experiment
1. The questions which were on strips of paper were stapled to the back of each consent
form (see appendix i)
2. The consent forms were then shuffled to randomize the questions given to the
participant. (as the question determines the condition to which the participant would
fall under)
3. The participants were handed the randomized consent forms and were asked to sign
them if they were willing to participate
4. If the participant signed the form, directions were provided on how to do the activity,
5. The participant was asked to turn the paper over and answer the question attached at
the back in 10 seconds.
6. The consent forms were collected and a debrief form(see appendix iii) was handed to
the participants
6
Analysis
Anchor condition
Median
of
tendency(MOCT)
central 43.0
Median of dispersion(MOD)
3
Control condition
42.0
5.25
The data we collected was ordinal (Appendix iv), categorical data with a set order or scale.
Hence the median of the collected data used as the MOCT while the semi interquartile range
was used as the MOD. As seen above the MOCT for the participants in the anchor condition
is slightly greater than that of the participants of the control condition with values 43.0 and
42.0 respectively. However, In terms of the MOD, control condition participants received a
value greater than the anchor condition participants with values 5.25 and 3 respectively. This
suggests that even though the anchor condition was introduced to a numerical anchor it had a
near negligible impact on the relevant estimates of age when compared with the control
condition which did not receive a numerical anchor. However, it can be implied through the
values of the MOD of both conditions that the reason the variability of estimates of anchor
condition is less than the control condition because of the influence of the anchor as it
suggests that the estimates provided by the participants in the anchor condition are pulled
towards a certain value rather than the control condition participants thus causing minimal
dispersion of estimates.
7
A Mann-Whitney U test indicated that there was no significant difference between age
estimates for the anchor condition (Median = 43) and the control condition (Median = 42), U
= 73, p = 0.28774 (Appendix v). To reject the null hypothesis p ≤ 0.05. However, the p-value
arrived upon through the test is much larger than the required range of p values to reject the
null hypothesis. Thus it can be concluded that there is no significance between the control
condition and the anchor condition, which suggests that the anchor provided to the
participants of the anchor condition has had almost no effect on their estimations of the age of
death of Chadwick Boseman thus confirming our null hypothesis and disproving our research
hypothesis.
8
Evaluation
Even Though our results do not agree with the original study it can be assumed that this
might have been a result of using a famous figure well known among the participants which
resulted in the anchor value being inefficient as the participants would have a stronger
understanding of him, rather than the participants of the original study who might have not
known Mahatma Gandhi. However, we might have been able to observe the effects of the
anchor if the anchor value was more plausible ( 60-70). Regardless, it would be an interesting
extension to observe the extent to which an anchor could influence someone's decision
making based on the level of knowledge/understanding the person has regarding the specific
situation.
One strength received through the use of independent measures design is that we were able to
reduce order effects which could have arisen if we had allowed each participant to take part in
both control and anchor conditions because that will make the participants incline to give the
same age estimates for both questions due to the similarity of the questions. However, one
constraint we faced during the experiment was the lack of participants we were able to
recruit(28). Since the design is independent measures this could have led to possible
individual differences of participants which could have influenced the results. Although this
issue could have been solved through the use of a different research design such as repeated
measures, we would arrive at the same issue stated above where the participants will give the
same age estimate for the 2 questions in the 2 conditions due to their near similarity.
Opportunity sampling turned out to be extremely beneficial to us in terms of the efficiency in
which participants were able to be recruited, considering the time constraints we had to
experiment (1 hour), however, this sampling technique could have led to selection bias which
could have led to the unreliability of our results even though steps were taken to limit them
(mentioned in exploration). This bias would have been avoided if volunteer sampling was
used instead as this would prevent us from choosing participants thus preventing our implicit
biases from influencing our selection of participants.
The procedure carried out during the experiment was sound, especially the standardized
directions as the participants were able to experiment without uncertainty and hesitation.
However, one major drawback of the experiment was the time allocated for the participants to
answer the question( 20 s) as it might have given enough time for the participants to switch
from system1 thinking to system2 thinking thus causing them to answer the question based
on careful thought rather than heuristics thus causing the introduced anchor(in the anchor
condition) to be ineffective. Therefore it would have been beneficial to limit the time to
answer the question to 10s.
9
To conclude, the p-value received through the Mann-Whitney U test carried out on the results
gained by the experiment is much greater than 0.05 suggesting that the control condition
estimates and the anchor condition estimates have no significance, therefore, retaining the
null hypothesis and hence justifying that the difference between the estimated age of death
given by the participants subjected to the high plausible anchor and the participants in the
control condition is insignificant.
10
Work cited
cherry, K. E. N. D. R. A. (2021). How Heuristics Help You Make Quick Decisions.
VerywellMind.https://www.verywellmind.com/what-is-a-heuristic-2795235#:%7E:text=A%20heuristi
c%20is%20a%20mental,their%20next%20course%20of%20action.
Strack, F., & Mussweiler, T. (1997). Explaining the enigmatic anchoring effect: Mechanisms of
selective accessibility. Journal of Personality and Social Psychology, 73(3), 437–446.
https://doi.org/10.1037/0022-3514.73.3.437
Independent
Groups
|Psychology.
(2018).
Tutor2u.
https://www.tutor2u.net/psychology/topics/independent-groups-design#:%7E:text=Independe
nt%20groups%20design%20is%20an,each%20condition%20of%20the%20experiment
Mcleod, S. (2019). Sampling Methods | Simply Psychology. SimplyPsychology.
https://www.simplypsychology.org/sampling.html#:%7E:text=Opportunity%20Sampling&te
xt=An%20opportunity%20sample%20is%20obtained,coming%20out%20of%20the%20librar
y
Selection
bias.
(2014).
Institute
for
Work
https://www.iwh.on.ca/what-researchers-mean-by/selection-bias
and
Health.
Blog, F. (2019, October 10). What is Ordinal Data? Examples, Variables & Analysis. Formpl.
https://www.formpl.us/blog/ordinal-data
Stangroom, J. E. R. E. M. Y. (2021). Social Science Statistics. Socscistatistics.
https://www.socscistatistics.com/
11
Appendix i - form of consent
Study Title: Observation Of The Effects Anchoring Bias Through The Replication Of Strack &
Mussweiler
Experimenter(s): DHANA KODITUWAKKU, JAMES ILLINGWORTH, RANDEV RANJIT
Description Of The Experiment: You will be required to guess the age of a famous individual.
To participate in this research study, you must give informed consent. By signing these
informed consent statements, you indicate that you understand the nature of the research
study and your role in that research and agree to participate in the study. Please consider the
following points before signing:
-
-
I understand that I am participating in psychological research;
I understand that my identity will not be linked with my data and that all of the
information I provide will remain confidential;
I understand that I will be provided with an explanation of the research I participated in
and will be given the name and telephone number of an individual to contact if I have
questions about the study.
I understand that participation in research is not required, is voluntary and that after
any individual research project has begun, I may refuse to participate further without
any obstruction.
By signing this form, I am stating that I am 16 or older and that I understand the above
information and consent to participate in this study being conducted at UWC Dover.
Signature:____________________________________________
Date:________________
12
Appendix ii-questions of the conditions
Control condition question:- At what age did Chadwick Boseman die?
Anchor question condition:- Did Chadwick Boseman die before the age of 90?
13
Appendix iii-Debrief
14
Appendix iv-Raw data
control
anchor
32
30
34
37
34
39
37
40
38
41
42
42
42
43
42
44
43
44
45
45
47
46
47
48
50
60
15
Appendix v-Mann-Whitney U test results
16
17
18
Download