Eliminating Gender Bias in Computer Science Education Materials

Eliminating Gender Bias in Computer Science
Jared O'Leary
Download

In this episode I unpack Medel and Pournaghshband’s (2017) publication titled “Eliminating gender bias in computer science education materials,” which examines three examples of “how stereotypes about women can manifest themselves through class materials” (p. 411)

  • Welcome back to another episode of the

    CSK8 podcast my name is jared o'leary

    every week the podcast alternates

    between an interview with guests or

    multiple guests and a solo episode where

    i unpack some scholarship

    this week is going to start a launch of

    a little mini series on

    gender bias in computer science

    education and i'm starting with a paper

    titled eliminating gender bias and

    computer science education materials

    this paper was written by paula medell

    and vahab pornashband

    in my apologies if i mispronounce any

    names in the show notes you can find

    a link directly to this particular paper

    and if you click on the author last

    names you'll be able to

    read some of their other works as it

    will take you to their google scholar

    profile

    all right so here's the abstract for

    this particular paper

    quote low female participation in

    computer science is a known problem

    studies reveal that female students are

    less confident in their cs skills and

    knowledge than their male counterparts

    despite parallel academic performance

    indicators while prior studies focus on

    limited

    apparent factors causing this lack of

    confidence our work is the first

    to demonstrate how in cs instructional

    materials may lead to the promotion of

    gender inequality we use a

    multidisciplinary perspective to examine

    profound but often subtle portrayals of

    gender bias within the course materials

    and reveal

    their underlying pedagogical causes we

    examine three distinct samples of

    established cs teaching materials and

    explain how they may affect female

    students

    these samples while not a complete

    display of all gender inequalities in cs

    curriculum

    serve as effective representation of the

    established terms a male-centered

    representation

    imagery and language that may promote

    gender inequality

    finally we present easily implementable

    alternative gender equitable approaches

    that maximize gender inclusion

    end quote that abstract does a pretty

    good job of summarizing the paper itself

    now if i

    summarize this into a single sentence i

    would say that this study examines three

    examples of quote

    how stereotypes about women can manifest

    themselves through class materials

    end quote that quotes from page 411 so

    the paper begins with a short

    introduction that kind of talks about

    some of the ways that stereotypes can

    negatively harm

    women in particular they talk about how

    it affects confidence

    in relation to computer science so even

    when women are performing just as

    well as men they're consistently having

    lower rates on their confidence rating

    in fields like computer science so after

    this short introduction

    the paper talks about the different

    materials that they analyze so in

    particular they analyze

    how names are represented within cs

    materials

    they talk about imagery within cs

    materials and then they also talk about

    pronouns within cs materials

    so in the paper it first begins by

    describing a common problem

    in cryptographic protocols so in this

    particular example

    it's basically people sending messages

    to each other and showing how people can

    intercept those messages or change them

    or whatever so what they did for the

    analysis is they took

    the names of the people within this

    particular example

    and they associated it with either a

    positive a negative or a neutral stance

    so for example eve was labeled as

    an eavesdropper and they were

    intercepting messages between two other

    people

    and able to read those messages that

    were being sent

    like through the internet so in that

    instance eve

    being a female name being labeled as an

    eavesdropper and doing something

    negative

    that one would receive a negative

    similarly

    mallory who was associated with a

    man-in-the-middle attack

    was also associated negatively however

    males

    tended to be associated with more

    positive things so like walton was the

    protective warden

    now in these examples they tended to

    have whatever the first letter of the

    name was

    associated with whatever it was that

    they were doing whether it was

    eavesdropping

    or warden or whatever so e for

    eavesdropping

    w for warden etc now the authors are

    arguing well

    you could have used any kind of name for

    this whether it be

    like a general neutral name such as alex

    or chris just as easily as you could

    have associated

    positive traits to females and negative

    traits to males

    but in general what they were finding is

    females were associated with a negative

    and males were associated with positive

    so one more example of this

    that is not only example of some

    gender biases but also ableism is the

    sybil attack

    so the sybil attack was previously known

    as pseudo-spoofing

    and it's some kind of an attack where

    quote identities are forged to support a

    reputation system and peer-to-peer

    networks end quote

    that's from page 412. so an example of a

    civil attack might be something like

    creating a bunch of false accounts and

    giving false reviews

    on a service or creating bot accounts

    on like something like twitter and then

    promoting things

    or arguing against things so you're

    making it seem like there's this

    mass amount of people who are asking for

    or recommending or arguing against

    something

    but really it might just be one person

    or a small

    number of people who are engaging in

    what is commonly referred to as a sybil

    attack

    now the reason why this particular

    example is brought up in this paper

    is that quote the name was inspired by

    the book sybil about the treatment of a

    woman diagnosed with disassociative

    identity disorder

    as a result of physical and sexual abuse

    the representation of a mentally ill

    woman

    as the field standard term for an

    attacker is not only insulting but

    harmful

    by projecting negative stereotypes about

    women unquote from page 412. and by the

    way

    disassociative identity disorder was

    formally referred to as multiple

    personality disorder in case you're

    unfamiliar with it there's just a

    clarification

    basically the same idea new term so with

    these examples

    whether it be the civil attack or eve

    the eavesdropper

    in the cryptography example the authors

    are basically arguing we need to analyze

    how we're portraying different genders

    within the materials that we're using

    now here's the reason why so here's a

    quote from page 412 quote

    by comparing characters with positively

    or negatively associated roles

    we found clear gender discrepancies

    there are more female characters than

    males

    however this does not indicate fair

    inclusion in fact

    of the four characters with positive

    connotation only one is female

    by comparison of the nine total negative

    roles six are female

    and three are male thus of eight female

    associable characters

    less than thirteen percent of them are

    good compared to fifty percent of

    associable male characters

    in quote okay so after kind of laying

    down some evidence that supports the

    idea that hey there's some gender bias

    here in how we're using these names and

    the associations we're giving to them

    they talk about what are some ways that

    we could be more equitable

    so one potential solution is to replace

    names with gender neutral names however

    the authors

    argue that there are still associations

    with particular genders for different

    names

    so for example the name alex if i have

    a friend who identifies as female named

    alex

    i might associate alex with female more

    so than i do with alex

    with male or non-binary but if i have a

    friend named alex who identifies as male

    i might associate it more with male

    so instead of using gender neutral names

    the authors actually recommend

    using animals so for example the

    eavesdropper could be the owl

    and instead of going with sybil we could

    say chameleon

    because chameleons change colors and

    assumes varying identities

    now the authors do say that quote due to

    the universal nature of animal

    representations

    educators from different cultural and

    language backgrounds can use this method

    to teach their students in a relatable

    way

    end quote while i understand what

    they're trying to say i disagree

    so some cultures view animals

    differently than other cultures

    for example cultures that use some of

    the stories from the bible about snakes

    being

    sneaky and subversive and manipulative

    and whatnot

    might differ than other cultures that

    represent snakes in a positive light

    so as an example some cultures actually

    view snakes within a sacred role

    or as representation of changes in

    cycles

    by the shedding of skin in other words

    not negatively so that's my one

    small minor disagreement with the what

    they're indicating

    however in general this recommendation

    of using animal characters instead of

    people names makes sense in relation to

    the gender biases that they're trying to

    avoid okay so the next particular

    example that they talk about

    is imagery that is used within materials

    so for example there's an image named

    lena

    that is often used for image processing

    examples and it is actually

    an image of a woman from a playboy

    magazine

    and they cropped it so that it's that

    person's bare shoulders

    and above and this image is frequently

    used in presentations publications etc

    when discussing image processing

    examples now here's a quote from page

    quote such imagery objectifies women by

    projecting stereotypes that emphasize

    their physical appearance rather than

    their mental values

    objectifying imagery affects women's

    confidence and therefore academic

    performance in two ways

    deteriorating their perceptions of self

    and lowering others perceptions of them

    end quote now the authors point out that

    some people have actually flipped the

    image so it was like a

    exposed version of a male in a similar

    way

    in that they cropped it up the shoulders

    and above and had a male model

    as the example but the authors argue

    that this is still an example of

    objectifying members of a different

    gender so some people have recommended

    well instead of using sexualized imagery

    how about we instead

    have positive imagery of different

    genders so for example having a picture

    of a woman holding a trophy or a woman

    in leadership

    but the authors actually recommend

    instead of using pictures of people

    to instead use pictures of monuments

    such as like pyramids

    or architecture or things like that the

    authors argue that this can

    help eliminate gender biases and that

    when you need to use

    facial images as examples so for example

    if you are creating materials that's

    talking

    about facial recognition and you need a

    picture of a face

    then recommend that using some kind of a

    picture that empowers people rather than

    objectifies them

    now i totally agree about the point

    of avoiding objectification of genders

    but i just want to point out that

    there's some debate about whether or not

    this is a form of objectification or a

    form of empowerment

    i'm personally not well-versed enough in

    that kind of scholarship in that area of

    study to be able to explain more nor am

    i

    a woman or identify as it identify as a

    non-binary by the way but i completely

    agree that

    we should steer away from sexualized

    imagery in course materials

    as i think that is particularly

    problematic or at least can be

    depending on the context and whatnot so

    the third area that they're analyzing

    is language so in particular they talk

    about examples of pronoun use

    so using only he only she

    he or she or the singular use of

    they while some people prefer to use he

    or she

    or he and she when referring to groups

    of people

    or just some like anonymous pseudo

    person in some kind of example

    the authors instead recommend that you

    use the singular

    they pronoun to refer to an unspecified

    gender

    now as a non-binary individual the

    pronoun they is the pronoun that most

    aligns with my own gender identity but i

    personally

    don't have a preference so you can use

    he she they with me

    so it doesn't really matter to me

    however if you use they

    it at least moves outside of the binary

    it does not put a particular gender

    within a positive or negative light it's

    more ambiguous

    and it again accounts outside of the

    gender binary

    so that recommendation totally relates

    to me and i highly recommend it

    now as educators some of the things that

    we need to think about is the ways that

    we speak with our students so it's not

    just in the materials that we submit

    so not just the names not just the

    imagery not just

    the pronouns on the assignments that we

    give but how we actually speak to people

    so for example a lot of youtubers will

    use the what's up guys or

    whatever other intro and it's that use

    of guys

    that can become problematic so i know

    some teachers who will avoid that and

    say

    good morning boys and girls or whatever

    something like that but again that then

    promotes the

    binary assumption with genders and makes

    non-binary trans individuals

    uncomfortable or at least can so we can

    avoid that by saying

    like good morning blank just end there

    or

    something else or using some kind of

    other group identity so as an example of

    this when i was originally creating

    the videos for boot up where i walk

    through step by step how to do stuff in

    scratch that kids are going to use

    it was very intentional with the opening

    line that i

    started with so every single video i

    start with welcome back

    fellow coders so it was a very

    intentional set of four words

    i went with coders because it's gender

    neutral

    it's also saying hey you are a coder

    you are a programmer you can do computer

    science and by saying fellow

    and saying hey i can program you can too

    so it was trying to avoid any kind of

    gender associations

    basically saying hey welcome back to

    this video i'm a coder you're also a

    coder

    although some people might argue that

    i'm thinking way too much about word

    choice

    it can have a huge impact so speaking of

    impact the authors

    took their suggestions and they actually

    implemented it into

    an experimental group that received the

    treatment i.e the

    replacements of people names with

    animals and the

    imagery of the sexualized woman with

    architecture or structures

    and then replacing the pronouns with the

    singular they so that's the group that

    received all those

    treatments and then a control group

    which was a class that

    just had the normal cs materials with

    these gender biases in them

    now what they ended up finding is that

    there was improvement for

    female students in terms of their

    confidence while mal students

    in either the experimental or the

    control group did not have any kind of

    statistically significant change in

    their confidence so it did not

    negatively impact them

    but it possibly impacted females in

    terms of their confidence

    so again as a quick summary of the paper

    itself they looked at

    names in course materials they looked at

    imagery and course materials and they

    looked at pronouns and course materials

    their overall recommendations were to

    avoid names and instead use animals

    to avoid imagery of sexualized genders

    and instead use

    something like a structure ideally a

    structure that is not like phallic-like

    or gendered and then to use the singular

    they

    instead of he or she for your pronouns

    as always at the end of these unpacking

    scholarships i like to share some

    lingering questions or thoughts

    or sometimes rants like a couple weeks

    ago so one question that i have that i

    honestly don't know and don't have an

    answer for is is the shift towards

    animals and monuments a form of

    dehumanizing computer science

    in other words are we taking the human

    aspects out of it are we making this

    technological thing

    even less human than it already can be

    at times and that

    i honestly don't know so i'm just kind

    of thinking out loud another question

    that i have is when creating or sharing

    materials with students

    what kind of demographic balances do you

    strive for

    so are you trying to demonstrate equal

    relationships match demographic

    proportions

    or are you leaning more toward

    marginalized identities to counter the

    trends

    so for an example related to gender and

    unlike the article i'm going to include

    non-binary within this so if you're

    trying to go for equal relationships are

    you going to have

    one-third female representation

    one-third non-binary representation and

    one-third male representation

    or if you're going for matching

    demographic proportions are you going to

    go for

    these are hypothetical numbers 50 female

    non-binary and 49 male representation

    and if you're going to go lean towards

    more marginalized identities to counter

    the trends

    so for example leaning towards

    representation and only 10

    male representation to counter balance

    the overabundance

    of males within cs materials in

    whichever direction that you end up

    going

    when might an approach like this

    unintentionally communicate messages

    that a certain demographic

    is not welcome within the cs community

    in other words does the pendulum then

    shift the other way

    so if we look at gender within cs and

    say well there's an over

    abundance of representation of males

    should we then shift the pendulums that

    we mainly focus on

    females non-binary representation does

    that then unintentionally say

    males are unwelcome in cs education and

    just like my

    question about dehumanizing i don't know

    again just thinking out loud

    it's something that i would love to see

    more research on and more conversations

    on

    within the field so my last question is

    not a question that is tied to this

    particular study

    but gender imagery analysis in general

    so the question is

    how might we as a field start engaging

    in conversations around gender without

    making assumptions about people

    now the reason why i say this is because

    i sat in a on a presentation once where

    somebody started playing a video of the

    classroom

    and the imagery within it and their

    commentary on the imagery was making

    assumptions about

    the genders that represented now if you

    looked at it there were a lot of male

    presenting individuals in there in terms

    of

    the ways that they were dressing in

    terms of their hairstyles etc

    and the comment was that this was a male

    dominated class however i would argue

    we actually don't know if that was a

    male-dominated class without actually

    asking

    the people within that imagery what we

    don't know is

    the class could have in fact been

    dominated by non-binary and trans

    individuals and we don't know until we

    actually

    do more than a surface level analysis of

    what we're seeing now i say this

    to say we should dive deeper into these

    gender discussions

    but also in recognition of the larger

    point was that yes

    cs is largely dominated by males

    wholeheartedly understand that

    so those are just some of my lingering

    thoughts related to the overall topic of

    this particular paper

    i enjoyed reading this paper and i enjoy

    these kinds of analyses

    so i highly recommend reading it if it

    also interests you

    again you can find it in the show notes

    if you enjoyed this particular episode

    please consider sharing it

    with a friend or colleague as it helps

    spread the word about cs education and

    research

    stay tuned next week for another

    interview and the following week for

    another unpacking scholarship episode

    i hope you're all having a wonderful

    week and are staying safe

Article

Medel, P. & Pournaghshband, V. (2017). Eliminating gender bias in computer science education materials. In Proceedings of the 2017 ACM SIGCSE Technical Symposium on Computer Science Education (SIGCSE '17). Association for Computing Machinery, New York, NY, USA, 411–416.

Abstract

“Low female participation in Computer Science is a known problem. Studies reveal that female students are less confident in their CS skills and knowledge than their male counterparts, despite parallel academic performance indicators. While prior studies focus on limited, apparent factors causing this lack of confidence, our work is the first to demonstrate how, in CS, instructional materials may lead to the promotion of gender inequality. We use a multidisciplinary perspective to examine profound, but often subtle portrayals of gender bias within the course materials and reveal their underlying pedagogical causes. We examine three distinct samples of established CS teaching materials and explain how they may affect female students. These samples, while not a complete display of all gender inequalities in CS curriculum, serve as effective representations of the established trends of male-centered representation, imagery, and language that may promote gender inequality. Finally, we present easily implementable, alternative gender equitable approaches that maximize gender inclusion.”

Author Keywords

Gender, Diversity, Confidence, Gender Equitable

My One Sentence Summary

This study examines three examples of “how stereotypes about women can manifest themselves through class materials” (p. 411)

Some Of My Lingering Questions/Thoughts

  • Is the shift toward animals and monuments a form of dehumanizing CS?

  • When creating or sharing materials with students, what kind of demographic balances do you strive for?

    • Are you trying to demonstrate equal relationships (e.g., 1/3 female, 1/3 nonbinary, and 1/3 male), match demographic proportions (e.g., 50% female, 1% nonbinary, and 49% male), or lean more toward marginalized identities to counter trends (e.g., 70% female, 20% nonbinary, and 10% male)?

      • When might an approach like this unintentionally communicate messages that a certain demographic is not welcome in the CS community?

  • How might we as a field start engaging in conversations around gender without making assumptions about people?

Resources/Links Relevant to This Episode

More Content

Unpacking ScholarshipJared O'LearyGender, Curriculum, Bias, EquityComment