Example 5: Student work

IB Psychology Internal Assessment

[HL]

Investigating the effects of anchoring bias

Candidate Code:

Group Member Codes:

Word Count: 2198

Psychology teacher support material 1

 Example 5: Student work

Contents:

Introduction - 3
Aim
Variables
Hypothesis
Null Hypothesis
Exploration - 6
Design
Sampling
Controls
Materials
Analysis - 7
Descriptive Statistics
Inferential Statistics
Evaluation - 9
Discussion of Findings
Sample
Design
Procedure
Modifications
Bibliography - 13
Appendices - 14
Raw Data
Mann Whitney U Test
Brief and Debrief
Consent Form
Question sheets
Procedural Instructions for Group

Psychology teacher support material 2

 Example 5: Student work

Introduction

Thinking is the process of recalling previously obtained knowledge and using it to go about

your daily life, such as interpreting situations, making predictions, etc. Decision making can

be defined as the action of identifying different alternatives and selecting a specific one

depending on your preference or personal values. One theory investigating thinking and

decision making is the Dual Process Model, which follows the idea that there are ‘two basic

modes of thinking’ known as ‘System 1’and ‘System 2’. The intuitive means of thinking would

fall under system one, this is automatic and rather fast. System two, is the rational, conscious

mode of thinking. It allows us to effectively analyse the situations we encounter by interpreting

the experience logically. (Student.thinkib.net, 2020)

There are flaws that coincide with using system one to make decisions because of shortcuts

known as heuristics. Heuristics use past experiences and assumptions to speed up the

decision making process, however these lead to cognitive biases. Despite usually being

effective, some cognitive biases create errors that are methodical and predictable. (Vinney,

2019) Psychologists Amos Tversky and Daniel Kahneman in 1969 began to investigate these

biases that would ultimately win Kahnenman a nobel prize in 2002. This IA will investigate

anchoring bias, and proposes that we sometimes estimate or assume using predetermined

information that can alter our interpretations.

Tversky and Kahneman (1974) carried out a study in order to research how anchors can yield

different estimates that are biased towards the initial value. (Tversky and Kahneman, 1974)

They used a mathematical equation in order to assess how participants answers will differ

depending on the anchoring effect of the order of the numbers. Two independent groups of

high school students were given 5 seconds to answer a numerical expression that was written

on a black board.

Psychology teacher support material 3 3

 Example 5: Student work

The first group was asked to estimate the product of: 𝟏𝟏 × 𝟐𝟐 × 𝟑𝟑 × 𝟒𝟒 × 𝟓𝟓 × 𝟔𝟔 × 𝟕𝟕 × 𝟖𝟖

While the other group was given: 𝟖𝟖 × 𝟕𝟕 × 𝟔𝟔 × 𝟓𝟓 × 𝟒𝟒 × 𝟑𝟑 × 𝟐𝟐 × 𝟏𝟏

It is thought that to answer such questions quickly we take short cuts. The results showed the

median estimate for the ascending sequence was 512 whereas for the descending sequence

the median was 2,250. Despite both being incorrect as the answer is 40,320, there is a clear

difference in the average answer. Therefore, supporting the theory that we use anchors when

making decisions quickly. (Tversky and Kahneman, 1974)

Aim:

The aim of our study is to investigate thinking and decision making, specifically the role of

anchoring bias when estimating the product of a mathematical calculation under time

pressure. This is relevant research to everyday life as we are constantly using predetermined

knowledge to aid our judgement when we don't have time to fully analyse a situation,

unconsciously making immediate assumptions. Such as when the price of an item is high and

then the salesmen reduces it for you you are more likely to buy that item than if it just began

at that reduced price as you believe that you are getting a good deal for your money. This is

based on the initial information which may lead to erroneous judgment.

Independent Variable:

The independent variable manipulated was whether the groups were given the question in the

ascending (1 × 2 × 3 × 4 × 5 × 6 × 7 × 8) or descending (8 × 7 × 6 × 5 × 4 × 3 × 2 ×

1)condition.

Psychology teacher support material 4

4
 Example 5: Student work

Dependent Variable:

Our dependent variable was the estimated numerical value given by each

participant in answer to the question.

Research Hypothesis (H₁):

The estimated answers given by the participants in the descending multiplication

condition (8 × 7 × 6 × 5 × 4 × 3 × 2 × 1) will be significantly higher than the estimated answers

given by participants in the ascending condition (1 × 2 × 3 × 4 × 5 × 6 × 7 × 8).

Null Hypothesis (H₀):

There will be no difference between the estimated answers given by the participants in

the descending multiplication condition (8 × 7 × 6 × 5 × 4 × 3 × 2 × 1) in comparison to estimates

given by participants in the ascending condition (1 × 2 × 3 × 4 × 5 × 6 × 7 × 8).

Psychology teacher support material

5 5
 Example 5: Student work

Exploration:

Design:

For our study an independent measures design was used as we made use of two seperate

groups of participants where both groups were given a different condition of the independent

variable. (Alleydog.com, 2019) In our case one group was given the ascending calculation and

the other group were given the descending calculation. We chose to use this design as the

participants would have been able to guess the aim if we were to use a repeated measures

design.

Sampling technique and Choice of Participants:

Our sample consisted of 27 sixth form students, studying either IB or A Levels in an

International School, in Asia. There were 12 male students and 15 female students in total, all

over the age of 16. The ascending equation was given to the Economics A level class and

descending to the IB French class. We used opportunity sampling to select these participants,

as they were available and willing to fully participate in the experiment. We chose these

students as they were the only available classes due to time tabling restrictions. We ensured

all students were proficient in English and had taken Maths up to GCSE, or an equivalent. We

believed that this age demographic and level of maths was sufficient to make a reasonable

estimate for the maths question.

Controlled Variables:

We ensured that each participant was given an equal amount of time, 30 seconds, to answer

the singular question. We used this amount of time as we believed that 5 seconds in Tversky

and Kahneman's study was too short and would have caused undue stress. All participants

from each group were given the same brief in order to maintain uniformity. We asked all

participants to work individually to prevent collusion disrupting the results.

Psychology teacher support material 6 6

 Example 5: Student work

Materials Used:

The testing rooms for each participant were extremely similar therefore the surroundings were

not a confounding variable. There were also no distractions that one condition encountered

and not the other. We used the same question sheet and the same consent sheet for each

participant, similarly to Tversky and Kahneman’s (1974) study. The only difference was that

we handed out individual question sheets, unlike the original study which consisted of a

singular chalk board with the question on it. We decided to do this as we thought it would

decrease the impulse to collaborate with others, focusing participant’s attention on the

question.

7
Psychology teacher support material 7
 Example 5: Student work

Analysis:

Descriptive Statistics

We collected ratio data, therefore calculating the mean would be most appropriate. However,

we are replicating Tversky and Kahneman’s (1974) study, as they calculated the median, we

did the same. In appendix A there is a table of our raw data. There is clear variability in the

estimations for each condition. For example, despite having the highest product of 150,000,

the descending condition also has an estimated product of 82. Likewise, in the ascending

condition there were values of 46 and 48 given. This could support why calculating the median

is more suitable in this case to the outliers that are present in the raw data. This is because

the median will be able to disregard these outliers when calculating the average, making sure

they have no impact on the result.

Ascending Descending

Median 1,267 425

Mean 14,568 6,475

Standard Deviation 41,038 14,261

Psychology teacher support material 8 8

 Example 5: Student work

The bar chart above displays the median result for the descending conditions was noticeably

higher than the median estimated value for the ascending condition with the descending

equation being 425 and the ascending equation being 1,267.

Inferential Statistics

To calculate whether there was a significant difference between the estimated values of the

ascending and descending equations, a Mann Whitney U test was carried out as our

investigation did not meet the parametric assumptions required in order to carry out a T-test.

(Socscistatistics.com, 2019) From the calculation that was carried out the U value was = 72,

and the critical U value was 56 at p<0.05, meaning the result is not significant. Therefore we

cannot accept our research hypothesis and accept the null hypothesis. This is interesting as

both the calculated median and mean results, illustrated in the table above, show a difference

in the values given by participants which supported Tversky and Kahneman’s study. Our

replicated study showed no difference between the two anchors that would be produced from

Psychology teacher support material 9 9

 Example 5: Student work

the mathematical calculations. This meant that our study did not support our hypothesis,

Tversky and Kahneman's (1974) original study or the theory of dual processing.

Psychology teacher support material 10 10

 Example 5: Student work

Evaluation:

Discussion of Findings

From the data collected in our study, the participants in the descending multiplication condition

(8 × 7 × 6 × 5 × 4 × 3 × 2 × 1) did not produce a significantly different estimate product than

the ascending condition (1 × 2 × 3 × 4 × 5 × 6 × 7 × 8). Our investigation did not support the

dual systems theory, specifically system one theory of thinking and decision making, as the

anchors presented did not significantly affect the participant’s estimated products. System one

being intuitive, fast paced thinking that often takes short cuts assumed by previously obtained

knowledge, which could lead to anchoring bias. System two thinking consists of considered,

logically processed thoughts that produce justifiable, rational conclusions.

Sample

By using opportunity sampling we were able to obtain willing participants quickly and easily.

This made carrying out our experiment a much faster process than if we had used random

sampling for example. Furthermore, our sample was an age demographic of 16 or over, thus

old enough to sign the consent for. There was a mix of girls and boys from diverse

backgrounds. The limitation of this sample however was by not using random sampling to

allocate participants there is a possibility of sampling bias. In addition, we can only generalise

our results to a target population with the same characteristics as the participants of our

sample, such as middle class, international high school students.

Design

As we used an independent measures design, we were able to prevent participants from

guessing the aim which could have led to participant demand characteristics such as

attempting to change ones answers to support or not support the hypothesis. Furthermore, it

prevents order effect as the participants won't get bored by repeating a similar task numerous

Psychology teacher support material 11 11

 Example 5: Student work

times affecting our results. However, as the two groups were independent of one another we

were not able to give the descending group the ascending condition and vice versa. Thus

introducing the possibility of participant variability. Meaning there may have been extraneous

variables in the participant's characteristics that we are unaware of, which then manipulates

the dependent variables and were the factors that decided the two groups results.

Procedure

When carrying out our experiment we ensured we gave the same mathematics questions to

our participants in the ascending and descending conditions as Tversky and Khaneman

(1974). This was important as it meant we were able to complete a valid comparison of our

results and the ones obtained in the original study. In addition because we handed out

individual sheets of paper with the question we believe it lowered attempts of collusion

between participants. Unlike the original study where they used a chalkboard to present the

equations, letting participants know they were all being given the same question.

Modifications

When carrying out our experiment we decided to change the given amount of time in the

original study by Tversky and Kahneman. We changed the 5 second time limit to 30 seconds.

This was because we thought 5 seconds was too short a time scale for people to fully read

the questions and answer them. However, after changing this we found that 30 seconds was

too long as it caused some people to think they had long enough to work it out, and attempted

to do so logically, but in the end were unable to answer. This may have created undue stress

as well as meant that they were no longer using estimation to answer the question. In the

future I would give the participants 15 seconds and see if this was a more appropriate time

scale. Furthermore, it would be interesting to see the difference in results obtained if the

experiment was carried out in a more naturalistic setting. Such as, the price someone will

Psychology teacher support material 12 12

 Example 5: Student work

agree is reasonable to pay for an item in a shop. Whether this price might differ depending on

the starting value of the item.

From our study we are able to conclude that anchors in a maths equation do not affect the

estimated value given as an answer.

Psychology teacher support material 13 13

 Example 5: Student work

Bibliography

Alleydog.com. (2019). Independent Measures Design definition | Psychology Glossary |

alleydog.com. [online] Available at:
https://www.alleydog.com/glossary/definition.php?term=Independent+Measures+Design
[Accessed 20 Oct. 2019].

Socscistatistics.com. (2019). Mann-Whitney U Test Calculator. [online] Available at:

https://www.socscistatistics.com/tests/mannwhitney/default2.aspx [Accessed 15 Oct. 2019].

Student.thinkib.net. (2020). InThinking Student pages. [online] Available at:

https://www.student.thinkib.net/psychology/page/22794/thinking-and-decision-making
[Accessed 25 Feb. 2020].

Tversky, A. and Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases.
[online] Its.caltech.edu. Available at:
http://www.its.caltech.edu/~camerer/Ec101/JudgementUncertainty.pdf [Accessed 20 Oct.
2019].

Vinney, C. (2019). How Cognitive Biases Increase Efficiency (And Lead to Errors). [online]
ThoughtCo. Available at: https://www.thoughtco.com/cognitive-bias-definition-examples-
4177684 [Accessed 26 Oct. 2019].

Psychology teacher support material 14 14

 Example 5: Student work

Appendix 1

Raw Data:

Descending Condition Ascending Condition

150,000 1,600

5,498 25,000

1,267 720

581 420

3,120 178

129 1,537

3,900 50,000

20,000 10,000

200 300

1,100 150

312 430

82 220

3,200 46

- 48

Psychology teacher support material 15 15

 Example 5: Student work

Appendix 2

Psychology teacher support material 16 16

 Example 5: Student work

Appendix 3

Brief:
Good afternoon,

We are conducting an experiment for our Psychology IA, which involves a short math quiz.
The only material that you will need for this experiment is a pen. We will give you a consent
form. Please fill in this form before proceeding.

You will be given 30 seconds to solve the problem. The question is on the other sheet of paper.
You are not allowed to do any working out on the paper. Since we are using this data for our
coursework, if you do not know the answer please give an educated guess, do not leave any
spaces blank. You also cannot communicate and collaborate with those around you. Thank
you.

Please note that you have the right to withdraw at any point of the experiment if you don’t wish
to continue.

Debrief:
The experiment that was just conducted was a version of Tversky & Kahneman’s 1974 study.
Overall our aim was to look into anchoring bias and how it affects the estimation of numbers.
Group 1 was the descending condition (8 × 7 × 6 × 5 × 4 × 3 × 2 × 1) and Group 2 was the
ascending condition (1 × 2 × 3 × 4 × 5 × 6 × 7 × 8). Anchoring bias is when the first number
you see acts as the anchor for the whole answer.

For those that are curious, the answer to the problem is 40,320, regardless of ascending or
descending conditions.

If you would like to receive the results of the study, please keep in contact with us at any of
the following emails:

Psychology teacher support material 17 17

 Example 5: Student work

Appendix 4

Consent Form:

Dear Participant,

Our names are ___________

In order to complete the IB psychology course we need to conduct an experiment, we have chosen to
specifically investigate cognitive processing and the theory of thinking and decision making. To meet
the ethical requirements, it is compulsory for you to fill out the form below if you would like to participate
in our study. Please only complete the form if you are both fluent in english and over the age of 16. If
you have any concerns or queries please ask the researcher.

I _______________________________ , confirm I have read and understood the participant

information and will receive a brief and debrief about the investigation from a researcher. I have had
the opportunity to ask and have all my questions answered. I understand that any participant information
collected will remain confidential and all efforts will be made to prevent the release of my identity. I
understand that any of my data gathered by the researchers will remain anonymous if published or used
in the future. I am aware that participation is voluntary and that I am able to withdraw myself and any
data produced from my results, with or without explanation.

I agree to partake in this study.

Participant Signature: _________________________________________

Participant Name: ____________________________________________

Date: _______________________________________________________

Psychology teacher support material 18 18

 Example 5: Student work

Appendix 5

Question Sheets:

Please estimate as quickly as possible the value of the following:

1×2×3×4×5×6×7×8=

Please estimate as quickly as possible the value of the following:

8×7×6×5×4×3×2×1=

Appendix 6

Our Instructions:
1. A class of 7 participants is chosen and will constitute as the ascending group
2. The participants are briefed and are given the consent forms with the question sheets
respective of their group on the backside
3. They are told to be silent for the duration of the study and make no contact with each
other
4. It is ensured that the participants are able to see the timer
5. It is announced when the timer starts and stops
6. Their consent forms and answers are collected
7. The participants are debriefed
8. A separate class of 7 different participants will constitute as the descending group
9. Steps 2-8 are repeated with the second group

Psychology teacher support material 19 19