How Early Does the CS gender Gap Emerge? A Study of Collaborative Problem Solving in 5th Grade Computer Science

In this episode I unpack Tsan, Boyer, and Lynch’s (2016) publication titled “How early does the CS gender gap emerge? A study of collaborative problem solving in 5th grade computer science,” which investigates the potential impact of gendered groups on the quality of completed Scratch projects in an in-school computer science class for 5th grade students.

  • Welcome back to another episode of the

    CSK8 podcast my name is jared o'leary

    each week of this podcast alternates

    between an interview and an unpacking

    scholarship where talk about some

    research in relation to cs education

    this week's particular episode

    is a continuation of a discussion on

    gender related

    publications that are relevant to cs

    educators in particular this week's

    episode

    is unpacking the paper titled how early

    does the cs gender gap emerge

    a study of collaborative problem solving

    in fifth grade computer science

    this paper was written by jennifer sun

    christie elizabeth boyer

    and colin f lynch apologies if i

    mispronounce any names

    as always you can find links to this

    particular paper in the show notes by

    clicking on the article

    title or you can find more publications

    by other authors

    by clicking on their names and it'll

    take you to their google scholar profile

    alright so here's the abstract for this

    particular paper quote

    elementary computer science has gained

    increasing attention with the computer

    science education research community

    we have only recently begun to explore

    the many unanswered questions about how

    young students learn computer science

    how they interact with each other and

    how their skill levels and backgrounds

    vary

    one set of unanswered questions focuses

    on gender equality for young computer

    science educators

    this paper examines how the gender

    composition of collaborative groups in

    elementary computer science relates to

    student achievement

    we report on data collected from an

    in-school 5th grade computer science

    elective

    offered over four quarters in 2014-2015

    we found a significant difference in the

    quality of artifacts produced by learner

    groups

    depending on their gender composition

    with groups of all female students

    performing significantly lower than

    other groups

    our analyses suggest important factors

    that are influential as these learners

    begin to solve computer science problems

    this new evidence of gender disparities

    in computer science achievements

    as young as 10 years of age highlights

    the importance of future studies on

    these factors in order to provide

    effective

    equitable computer science education to

    learners of all ages end quote

    now if i were to summarize this

    particular study into a single sentence

    i would say that this paper investigates

    the potential impact of gendered groups

    on the quality of

    completed scratch projects in an

    in-school computer science class for 5th

    grade students

    all right so in the introduction of this

    paper the authors describe

    how there's a difference in gender

    within like

    post-secondary high school and even

    middle school classes however there was

    very little research at the time of this

    publication

    on the gender differences in elementary

    school the authors note

    that in particular math has shown some

    gender differences as young as first

    grade

    so considering that there are some

    differences in that grade level or age

    group

    the authors are going to investigate

    whether or not there are some

    gender differences within fifth grade

    students now before actually doing this

    particular study they did a pilot group

    with an after school class

    and they co-developed this class with an

    elementary school teacher and then after

    completing this pilot and doing some

    revisions and whatnot they then

    implemented this class across

    four different quarters which i'll talk

    a little bit more about so this

    particular paper is guided by the

    question

    how is gender composition of

    collaborative groups in elementary

    computer science related to student

    achievement

    that question's from page 388 and here's

    a

    two sentence summary of the findings

    quote we collected students as programs

    from one project in the fifth grade

    computer science course these programs

    were rated for quality

    and the results showed that all female

    groups produced significantly lower

    scored artifacts than groups with at

    least one male

    end quote from page 389 okay so the next

    section of the paper talks about some

    related work so they highlight some

    areas of elementary school

    interventions such as like using other

    platforms

    or computer science as well as some

    discussions around collaboration in

    elementary school

    now under that particular section on

    collaboration elementary school there's

    an interesting quote from page 389

    quote outside of computer science a

    study of elementary students is

    collaboration revealed that female to

    male relationships

    in subjects such as physical education

    reading language art

    social studies and art are not equal

    boys appear to benefit more from female

    to male pairings by developing their

    leadership skills and increasing their

    self-efficacy in problem solving

    whereas these pairings can reinforce

    gender stereotypes for girls

    end quote so that is a very interesting

    finding if you are engaging

    in paired programming you'll likely want

    to read some more research on this

    finding and to figure out why

    is it that this kind of pairing might be

    beneficial for male students but

    detrimental to female students maybe

    i'll find some particular papers to

    discuss this

    in future unpacking scholarship episodes

    all right so the next section of the

    paper talks about some context so the

    teacher who was doing this had no

    formal background in computer science

    and the class set this was occurring in

    was a 30-day 45-minute class time

    that occurred in a quarterly rotation

    so new students would appear every

    quarter the demographics were 53.1

    percent african-american 32.6 caucasian

    and 14.3

    hispanic latino native american asian or

    mixed race

    and 47.4 percent of students receive

    free or reduced lunch

    so those stats are on page 390 those

    were basically a verbatim

    restatement of them now in this school

    year throughout each of the quarters

    there were a total of 55 students

    in this elective class 16 of which were

    girls and 39 of which were boys there

    was no mention of non-binary so i

    actually don't know

    whether or not they investigated whether

    there were non-binary individuals in

    this class

    or at least people who identified as

    non-binary at the time because if you

    would ask me when i was this age group i

    would have identified as male but

    it wasn't until a few years ago i was

    able to finally find some

    language that described how i actually

    feel and identify

    as a non-binary individual so in this

    class these students completed three

    major projects two of them were scratch

    programming projects and one was a

    research project

    on computer science and society so this

    particular study

    looked at the first programming project

    so this particular project was to create

    a fairy tale

    in scratch the students were required to

    incorporate two instances of cause and

    effect from the fairy tale

    and two instances of user input where

    the user's input decided the path the

    story took

    in quote from page 390. all right so the

    next section of this particular paper

    takes a look at 18 pairs of groups

    four were all female four were male and

    female pairings

    and then 10 were all male then there's

    also one individual

    who didn't work with a group now you'll

    already notice that there is a

    quite a gender imbalance in this so the

    results are going to obviously be a bit

    skewed

    in terms of having a much smaller sample

    size of females

    students and a much larger sample size

    of male students but as with all studies

    that are like this

    follow-up studies are important to

    figure out is this generalizable

    and if so where and how and why all

    right so in this particular

    study the things that they're analyzing

    with the projects themselves

    were looking at the requirements like

    whether or not they actually

    had two instances of user input and

    cause and effect

    they also looked at the consistency so

    whether the project actually ran

    consistently so like resetting the

    sprites

    if if needed at the start of the project

    and then usability so whether there were

    some prompts to like guide the user to

    figure out how to

    engage with this story so they did

    weight this so that

    towards the requirements

    consistency and then 10

    was going towards the usability now

    here's a quote from page 390 that kind

    of summarizes the overall results

    quote the programs of all four

    all-female groups

    scored below the median and the scores

    were clustered together

    the scores of the female male groups

    were also relatively close to each other

    with their lowest scores being equal to

    the highest score of the all female

    groups and their highest score being

    slightly lower than most of the all male

    groups

    the all male groups displayed the widest

    range of scores with both the highest

    and lowest scoring groups

    out of all the categories however note

    that the all male group was also

    the largest with an n of 10. seventy

    percent of the all male groups scored

    zero point seven seven or higher end

    quote

    and to clarify the maximum score you

    could receive is a one so

    in other words seventy percent of all

    the male groups scored at least 77

    percent or higher

    out of the total possible score so on

    the next page on 390 here's a following

    quote

    quote the results revealed that none of

    the all female groups and 64

    of the at least one male groups achieved

    a high code score

    the difference was statistically

    significant with a p-value of 0.04

    we also examined the all-male group n of

    and at least one female and of eight

    groups separately in the split seventy

    percent of all male groups and

    twenty-five percent of at least one

    female group achieved high code scores

    and this difference was not

    statistically significant the p-value of

    they give a few examples of case studies

    so the first one that they give is of an

    individual and they point out that this

    particular individual

    scored the highest possible score and

    that they had

    experience participating in the after

    school pilot program that occurred

    in the spring before this school year

    the next case study that

    they talk about is a female to male pair

    and they found that there were some

    interpersonal collaboration and

    communication issues

    that kind of negatively impacted the

    overall process

    the next case study that they talk about

    is a female to female pair

    and how they had good interpersonal

    communication and collaboration because

    they were friends

    and although their code did run

    consistently and they did have some user

    prompts

    the actual conversations were incoherent

    so the

    fairy tale didn't really make sense

    according to the people who

    assessed this and the last group that

    they provide a case study for is a male

    to male pairing

    and although they worked well together

    they were kind of distracted by looking

    at other projects

    and they scored low compared to the

    other all-male groups all right so the

    next section of this particular paper is

    on the discussions and the limitations

    so the first section talks about

    supporting design and planning so

    they're basically saying that

    this process could go a lot better for

    partner groups and for just creating in

    general

    by engaging in some supportive processes

    for learning how to design

    and plan for creating a project so for

    example in your class you might engage

    in storyboarding or

    pseudocoding or you write out a rough

    outline or framework for what you want

    to happen and how it will happen

    with some of the code that you could use

    in whatever platform you're using

    or language this is especially important

    for

    large projects or even some smaller

    projects if kids are

    less familiar with programming so the

    next section

    under the discussion is on fostering

    collaborative skills and practices

    and they were discussing how it's

    important to try and identify some good

    collaborative

    habits with the students and how these

    collaborative skills can be not only

    beneficial

    for the projects that are being created

    in the computer science class but also

    just

    in life in general in career or leisure

    the next section that they talk about is

    investigating cognitive and social

    constructs so i'm going to read a quote

    from here because this is important to

    consider this from page 392.

    quote differences in performance at

    young ages is almost certainly related

    in part to influences that occur prior

    to the class itself for example

    there is a systematic gender difference

    between toys marketed to young children

    legos and robots are more likely to be

    marketed for boys

    while dolls and kitchen sets are more

    likely to be marketed for girls

    it is possible that the traditionally

    boy toys foster spatial and programmatic

    reasoning skills

    more than the traditionally girl toys

    and recent literature suggests a

    correlation between spatial reasoning

    and computer science performance

    in quote and the boy toys and girl toys

    were in quotes so this really relates to

    what i talked about two weeks ago in

    that unpacking scholarship episode

    where i talk about how technology

    positions and careers

    in malaysia are dominated by women

    statistically speaking

    so if you haven't listened to that

    episode or the other episodes related to

    gender i highly recommend

    checking those out i'll include links to

    those in the show notes all right and

    then this

    particular section of the paper finishes

    with the discussion on some of the

    limitations i'm going to skip those but

    basically it's talking about small

    sample size

    how the course changed over each one of

    the quarters so you can't really compare

    quarter one to quarter four

    because the teacher was continuing to

    refine and improve et cetera

    all right so the paper concludes with

    the discussion on feature work

    and here's a quote from page 393 quote

    classroom observations suggested that a

    difference did exist and results

    demonstrated that all female groups

    achieves

    significantly lower program quality than

    groups with at least one male

    analyses suggest that numerous factors

    play important roles

    including the extent to which students

    were willing to engage in design

    students as collaborative practices and

    other cognitive and social factors

    end quote so this quote kind of leads

    into what i'd like to talk about next

    which are some of my lingering questions

    or thoughts on this particular paper so

    one question that i have is how might

    the gender and confidence of a teacher

    impact learning so as an example if a

    female teacher

    expressed that they didn't know much

    because it was their first time

    doing programming or coding would it

    negatively impact female students's

    own perceptions of how they might be

    able to do in coding

    or computer science however if the

    teacher is female and they

    express confidence would that have

    increased any of the confidence in the

    female students

    and how would this change for male

    teachers or non-binary or trans teachers

    in other words what is the impact or

    potential impact that a teacher might

    have in relation to their self-efficacy

    or confidence

    with coding and teaching coding so if i

    was in the classroom and i was

    expressing some hesitation about

    doing something or learning something or

    teaching something students

    related to my different identities may

    have

    also expressed some hesitation or been

    like well if jared can't do it then i

    don't know if i can do it but i don't

    know now the question that i have is how

    would the findings have

    changed over time so the authors do

    mention that they analyze the first

    project but what would happen if you

    compare the first with the last project

    will there be any differences in amount

    of growth or in final product quality

    so for example if the female and female

    groups were scoring the lowest on the

    first project did they make the highest

    gains in terms of overall score

    on their final project or did they

    remain at the same

    overall scoring or just scale with a

    relative amount

    of increase as the other groups

    also how might the findings have changed

    if the groups had improved their

    collaboration skills

    the authors do mention that multiple

    groups didn't do so well

    and it might have been because they

    weren't really great at collaborating

    and working with each other

    so one thing we might need to consider

    ses educators this is not just about

    learning the content

    but it's also all about being able to

    facilitate collaboration

    with other students and learning from

    and working with peers

    another question that i have is how

    might the results change if we accounted

    for prior experience

    so for example accounting for prior

    experience with computers

    prior experience with creating stories

    prior experience playing video games

    prior experience coding like one of the

    case studies one of the students was

    engaging in the pilot program that was

    after school

    or even just experience going home and

    being able to actually

    continue to work on the project at home

    since scratch is online

    how might all of these factors impact

    the overall quality of the final product

    and what is missing from

    the final product that could be learned

    by actually chatting with the students

    like oh what didn't you include and why

    in other research that i read

    some of the researchers have mentioned

    that the final product was poor

    but it was because students ultimately

    ended up throwing away much of the work

    rather than leaving it in their project

    so they may have scored a higher score

    if they had kept it in but they

    ultimately didn't like it so they just

    removed it which gave the false

    impression that they didn't understand

    what they were supposed to do or were

    incapable of doing it or really they

    were they just didn't like the overall

    end product

    so they essentially erased their work

    the final question that i have is

    how would the results change for trans

    students so for example would we find

    similar results for trans masculine

    students and males

    and what about trans feminine students

    and females what about

    non-binary non-conforming genderqueer

    etc

    folks who kind of who exist outside of

    leaning towards

    one of the binaries that is often

    presented with gender so to be more

    concrete with what i'm saying

    if you had a female and female group a

    male and male group

    and a non-binary and non-binary group

    and we were to take the stats of this

    particular study

    would the female and females go their

    lowest the non-binary and non-binary

    score in the middle and the male and

    male score the highest

    if so why obviously this is all

    hypothetical but it's important to

    consider it's also important to consider

    that while these studies

    are helpful to think through these

    things we can't generalize and say

    oh well because of this now i know that

    my female and female pairings are just

    obviously going to score lower than my

    male to male pairings in the classes

    that i work with no we can't make that

    conclusion

    but we can look at this and use it as a

    heuristic and say hmm

    i wonder why that is and how can i

    improve upon

    the pairings in my class so as an

    example i personally preferred to allow

    kids to determine whether they wanted to

    work in groups or work on their own

    i was one of those kind of students who

    always preferred to work on my own even

    if it meant doubling my amount of work

    i just learned more that way and i

    enjoyed that process

    when students did want to pair with each

    other i made sure that they were getting

    along and that they were

    both learning from this experience so

    for example the

    article mentions every 15 minutes like

    switching between

    who's the driver and who's the navigator

    that can work

    but some students prefer to be the

    driver and some students prefer to be

    the navigator as long as they can both

    understand what is going on that was

    fine with me and for me i also

    encouraged groups that were larger than

    just pairs as long as everybody could be

    actively engaged and demonstrate they

    all understand

    what was going on totally cool with me

    but those were just my approaches

    so you might have some other approaches

    all right so that concludes this week's

    particular episode

    in this little mini series on gender i

    will certainly talk about it more in

    future

    interviews and unpacking scholarship

    episodes but i hope this little mini

    series gave you some things to consider

    in relation to

    gender and cs education if it did please

    consider sharing one of these episodes

    with somebody you know

    and thank you so much for listening to

    this particular episode i hope you have

    a safe and wonderful week

Article

Tsan, J., Boyer, K. E., & Lynch, C. F. (2016). How early does the CS gender gap emerge? A study of collaborative problem solving in 5th grade computer science. SIGCSE 2016 - Proceedings of the 47th ACM Technical Symposium on Computing Science Education, 388–393.


Abstract

“Elementary computer science has gained increasing attention within the computer science education research community. We have only recently begun to explore the many unanswered questions about how young students learn computer science, how they interact with each other, and how their skill levels and backgrounds vary. One set of unanswered questions focuses on gender equality for young computer science learners. This paper examines how the gender composition of collaborative groups in elementary computer science relates to student achievement. We report on data collected from an in-school 5th grade computer science elective offered over four quarters in 2014-2015. We found a significant difference in the quality of artifacts produced by learner groups depending upon their gender composition, with groups of all female students performing significantly lower than other groups. Our analyses suggest important factors that are influential as these learners begin to solve computer science problems. This new evidence of gender disparities in computer science achievement as young as ten years of age highlights the importance of future study of these factors in order to provide effective, equitable computer science education to learners of all ages.”


Author Keywords

K-12, elementary, gender diversity, collaboration


My One Sentence Summary

This paper investigates the potential impact of gendered groups on the quality of completed Scratch projects in an in-school computer science class for 5th grade students.


Some Of My Lingering Questions/Thoughts

  • How might the gender and confidence of a teacher impact learning?

  • How would the findings have changed over time?

  • How might the results change if we accounted for prior experience?

  • How would the results change for trans students?


Resources/Links Relevant to This Episode



More Content