Impostor Syndrome in Final Year Computer Science Students: An Eye Tracking and Biometrics Study

Presented at: The 18th International Conference on Augmented Cognition (AC ‘24)

Date of Conference: July 2024
Conference Location: Washington, DC

The preprint is available at: https://arxiv.org/abs/2404.10194

1. Alyssia Chen, Carol Wong, Katy Tarrit, and Anthony Peruma
S167: Applications of Augmented Cognition
HCI INTERNATIONAL 2024
Impostor Syndrome in Final Year
Computer Science Students: An
Eye Tracking and Biometrics Study

2. Introduction
01.

3. Impostor Syndrome
A psychological phenomenon
characterized by persistent self-doubt
of intellect, skills, or accomplishments
and feelings of fraudulence or
inadequacy despite evidence of one’s
competence and accomplishments.

4. Impostor Syndrome & CS
Students
57% of CS undergrads exhibit frequent
impostor feelings (Rosenstein et al.)
Higher prevalence among female
students(71%) vs. male students (52%)
Impostor syndrome persists even after
gaining experience and employment
(Thayer & Ko)
There is limited workon how impostor
syndrome impacts (final year)CS
students

5. Goal & Research Questions
Goal: This study aims to understand how impostor
syndrome impacts code comprehension
RQ1: To what extent are final-year undergraduate
computer science students confident in their
program comprehension skills?
RQ2: How does impostor syndrome affect
cognitive processes involved in comprehending
code?

6. Study Design
02.

7. Study Design
5 Java Code Snippets:
●Array size manipulation
Determine the output
●Bubble sort algorithm
Identify the error line
●Recursion
Determine the output
●String analysis
Determine the output
●Palindrome check
Identify the error line
Gazepoint GP3
HD Eye Tracker
Gazepoint
Biometrics Device
(Heart Rate)
15 Final-Year CS
Students
(9 male; 5 female;
1 non-binary)
Device Calibration
& Trial Run
Pre-Questionnaire
(Demographics)
Post-Questionnaire
(Clance IP Scale)
Code Comprehension
Activities

8. Data Analysis
Captured/Analyzed Metrics:
●Eye tracking metrics:
○Areas of Interest (AOIs)
○Time spent on each AOI
●Heart rate data
●Impostor Phenomenon
Characteristics (IPC) levels
●Correctness of answers
●Time spent on tasks
Per Participant:
●Average heart rate
●Total time spent on a
snippet
●Most/least looked at AOI
Across All Participants:
●Correct answers
●Average heart rate, time
spent, AOI

9. Widely used tool to measure impostor
syndrome
20-item questionnaire, range of multiple-
choice answers
Assesses:
●Fear of evaluation
●Fear of not being able to repeat success
●Self-doubt in abilities
Each answer mapped to a numerical value
from 1 to 5, respectively.
“I have often succeeded in
programming tasks, even
though I was afraid that I
would not do well before I
started working on it.”
Clance Impostor Syndrome

10. Results
03.

11. Impostor Syndrome Distribution
●Few: 3 participants (all male)
●Moderate: 5 participants (2 female, 3 male)
●Frequent: 5 participants (2 female, 3 male)
●Intense:2 participants (1 female, 1 non-binary)
Gender Differences
t-test of scores between males/females →statistically
significant (p < 0.05)
●males (M= 50.56, SD = 12.32)
●females (M = 68.80, SD = 12.28)
Industry Experience
1-way ANOVA tests of scores between varying software
industry experience and Java experience = p > .05 for both
To what extent are final-
year undergraduate
computer science students
confident in their program
comprehension skills?
Research Question 1
Female students have a
higher characteristics of
impostor syndrome

12. Impostor Syndrome and Code Comprehension
For each code snippet, we examine the following
between each IPC level
●Average heart rate
●Total time spent on a snippet
●Most looked at code AOI
●Correct answers
Then combine results for all code snippets for an
overall analysis.
Research Question 2
How does impostor
syndrome affect cognitive
processes involved in
comprehending code?

13. Code Snippet #1 -Array Size Manipulation
Only 1 participant(Moderate IPC)
answered correctly
Heart Rate:
Lowest median: Few IPC group
Highest median: Moderate IPC group
Time Spent:
Shortest: Few IPC group
Longest: Intense IPC group
AOI Focus:
Intense IPC: If statements

14. Code Snippet #2 -Bubble Sort
Only 2 participant(Frequent IPC)
answered correctly
Heart Rate:
Lowest median: Moderate IPC group
Highest median: Intense IPC group
Time Spent:
Shortest: Few IPC group
Longest: Intense IPC group
AOI Focus:
Intense, Frequent & Few IPC: Method
declaration
Moderate: Variable declaration

15. Code Snippet #3 -Recursion
All Few IPC answered correctly
No Intense IPC answered correctly
Heart Rate:
Lowest median: Frequent IPC group
Highest median: Intense IPC group
Time Spent:
Shortest: Few IPC group
Longest: Intense IPC group
AOI Focus:
AOI focus varied across IPC levels

16. Code Snippet #4 -String Analysis
7participantsanswered (3 Few; 3
Moderate & Frequent; 1 Intense)
Heart Rate:
Lowest median: Few IPC group
Highest median: Moderate & Intense
Time Spent:
Shortest: Few IPC group
Longest: Intense IPC group
AOI Focus:
Intense: Method declaration
Others: Variable assignment

17. Code Snippet #5 -Palindrome
8 participantsanswered (3 Few; 1
Moderate, 4 Frequent; 0 Intense)
Heart Rate:
Lowest median: Few IPC group
Highest median: Intense ICP group
Time Spent:
Shortest: Frequent IPC group
Longest: Intense IPC group
AOI Focus:
All groups spent most time on method
declarations

18. Research Question 2
How does impostor
syndrome affect cognitive
processes involved in
comprehending code?
Studentswithhigherimpostersyndromewere
associatedwithspendinglongertimeon
snippetsandwerelikelytoincorrectlyanswer
thequestionrelatedtoeachcodesnippet.
Acrossallcodesnippets,thosewithhigher
impostersyndromeweremorelikelytolookat
theMethodCallandMethodDeclarationcode
categories

19. Threats &
Conclusion
04.

20. Threats to Validity
We used mid-range research-grade
devices
Prior studies have used the same
Gazepoint Eye Tracker
w/Biometrics Kit
Controlled setting to focus on the
effects of imposter syndrome on
code comprehension
Followed established guidelines to
set up the eye-tracking experiment
Controlled Environment
Exploratory Study
Recruiting participants for in person
studies is challenging
Sample Size &
Single Institute
Concepts participants would have
encountered in academia and
interviews
Simplified Code

21. Conclusion
Female ↑IPC | Male ↓IPC
The male students showed
lower levelsof imposter
syndrome compared to the
females
↑IPC ↑Time Spent ↓Correct
Higher levelsof IPC were
associated with increased duration
of time spent on a snippet and
incorrect answers
↑IPC ↑Method Calls
Those with higher levelsof IPC are
more likely to spend more time on
method declarations and method calls

22. Thank you!
Alyssia Chen Carol Wong Katy TarritAnthony Peruma