Reinforcing Diversity Company Policies: Insights from StackOverﬂow

Developers Survey

Karina Kohl Silveira

, Soraia Musse

, Isabel Manssour

, Renata Vieira

and Rafael Prikladnicki

School of Technology, Pontif

ıcia Universidade Cat

olica do Rio Grande do Sul (PUCRS), Porto Alegre, Brazil

Keywords:

Software Engineering, Software Development, Diversity, Data Visualization, StackOverﬂow.

Abstract:

Diversity is being intensively discussed by different knowledge areas of society and discussions in Software

Engineering, are increasing as well. There are unconscious bias and lack of representativeness and when

we talk about characteristics as ethnicity and gender, to mention a few. How can tech companies support

diversity, minimizing unconscious bias in their teams? Studies say that diversity builds better teams and

delivers better results, among other beneﬁts. Cognitive diversity is linked to better outcomes, and is inﬂuenced

by identity diversity (e.g., gender, race, etc.), mainly when tasks are related to problem-solving and prediction.

In this work, we are interested in understanding the pain points in software engineering regarding diversity and

provide insights to support the attraction, hiring and retention policies for more diverse software engineering

environments. StackOverﬂow is a popular community question&answer forum, with a high engagement of

software developers. Yearly, they apply a survey, present straightforward results, and made the anonymized

results available for download. So, it is possible to perform additional analysis beyond the original ones. Using

data visualization techniques, we analyzed 2018 data implying insights and recommendations. Results show

that diversity in companies is not yet a conscious decision-making factor for developers assessing a new job

opportunity, and respondents from underrepresented groups tend to believe more they are not as good as their

peers. We also propose a discussion about the unconscious bias, stereotypes, and impostor syndrome and how

to provide support on that.

1 INTRODUCTION

More than ever, software development is a col-

laborative task. Software development teams are

built on people and, lately, the area is becoming

aware of the problem of underrepresented groups,

like gender, racial, cultural, etc. As mentioned in

Vasilescu (Vasilescu et al., 2014) previous work, gen-

der representation in Science, Technology, Engineer-

ing, and Mathematics (STEM) related subjects raises

the signiﬁcant attention of researchers and academics,

as well as of policy-makers, all noting a signiﬁ-

cant under-representation of women. Several compa-

nies are doing the challenge of embracing diversity

in the workforce as Google (Google, 2018a), Face-

https://orcid.org/0000-0002-2964-4681

https://orcid.org/0000-0002-3278-217X

https://orcid.org/0000-0001-9446-6757

https://orcid.org/0000-0003-2449-5477

https://orcid.org/0000-0003-3351-4916

book (Facebook, 2018), Microsoft (Microsoft, 2018),

and SAP (SAP, 2018).

Software Engineering demands well developed

problem-solving skills from developers. Considering

Agile Methodologies, for example, the Agile Mani-

festo (Mike Beedle, 2001), put people as critical as-

sets to better performance on development and deliv-

ery, by prioritizing ”individuals and interactions over

process and tools”. Roger S. Pressman (Pressman,

2010) says that the agile philosophy emphasizes in-

dividual competence (team members) combined with

group collaboration as a critical success factor for the

team. In other areas, the impact of diversity is also

an object of study. Menard et al. (Menard et al.,

2018) presents how to adequately address cultural di-

versity in organizations regarding practices of infor-

mation protection. F

orster (F

orster, 2018) shows the

effect of ethnic heterogeneity on electoral turnout in

Europe, offering insights into the role ethnic hetero-

geneity plays in political participation. Atadero et

al. (Atadero et al., 2018) discuss how to create an in-

Silveira, K., Musse, S., Manssour, I., Vieira, R. and Prikladnicki, R.

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey.

DOI: 10.5220/0007707901190129

In Proceedings of the 21st International Conference on Enterprise Information Systems (ICEIS 2019), pages 119-129

ISBN: 978-989-758-372-8

119

clusive environment in the ﬁrst years of engineering

courses in the light of efforts to broaden the partici-

pation of women and people of color in engineering

degree programs and careers. Software engineering

environments need to be safe for diversity and devel-

opers, managers and all other roles in these environ-

ments need to understand what are the factors that im-

pact work environments, and to help companies to re-

ﬁne their diversity and inclusion policies.

This paper aims to show our insights, based on an

analysis of StackOverﬂow Developers’ Survey data.

To achieve that, we are interested in exploring the fol-

lowing research question:

RQ. What companies working with software engi-

neering should focus to attract, hire and retain talents

on the specter of diversity?

To answer this question, we derived three other

questions:

RQ1. What is being considered by developers

when assessing job opportunities?

RQ2. What are their professional objectives for

the future?

RQ3. How conﬁdent developers are in their pro-

gramming skills?

Our analysis shows that diversity in the company

is not yet a conscious decision-making factor for a

developer assessing a new job opportunity except

for non-binary and transgenders. Also, respondents

that identiﬁed themselves as women, non-binary and

transgenders tend to doubt more their programming

skills and believe they are not as good as their peers

than the respondents identiﬁed as men. A discussion

about the unconscious bias, stereotypes, and impos-

tor syndrome and how to provide support on that is

provided in the Results and Discussions in Section 5.

The rest of the paper proceeds as follows. Sec-

tion 2 presents some background on how some big

players in the software engineering industry are deal-

ing with their diversity and inclusion strategies. Sec-

tion 3 shows some related work regarding data com-

ing from StackOverﬂow. Section 4 summarizes the

methods we used to analyze the data. Section 5

presents our results, Section 7 the threat to validity,

and Section 6 discusses them and their implications.

Section 8 concludes the paper.

2 BACKGROUND

Page (Page, 2007) says that we cannot tell whether

diversity is good or bad unless we ﬁrst know what

diversity is. They characterized diversity as the dif-

ferences in how people see, categorize, understand,

and go about improving the world, recognizing dif-

ferent dimensions of diversity: cognitive and iden-

tity. Cognitive Diversity is the difference between

how we interpret, reason and solve problems - how

we think. Identity Diversity is determined by afﬁlia-

tion with a social group as gender, culture, ethnicity,

religion, sexual orientation, etc. Identity diversity and

cognitive diversity often go hand in hand. People be-

longing to different identity groups, or with different

life experiences, also tend to acquire diverse cognitive

tools. Education, life experiences, and identity can all

contribute to cognitive diversity. How much of these

matters depend on the task. For identity diversity to be

beneﬁcial in a group, it must link with cognitive diver-

sity, and it happens when tasks are related to problem-

solving and prediction, so the identities translate into

relevant tools. Also, when the members of the group

have little or no preference diversity, and when they

get along with one another. In these cases, identity

diverse groups do perform better than homogeneous

groups (Page, 2017).

2.1 Efforts to have Diverse Workforces

To have diversity in the workforce is a challenge

embraced by several companies. Table 1 summa-

rizes the Gender distribution and Table 2 summarizes

Race/Ethnicity distribution in some huge technology

companies.

Google (Google, 2018a) is one of the technology

companies that are engaged in increasing their num-

bers in diversity and inclusion area. They believe

once their mission is to organize the world’s infor-

mation and make it universally accessible and useful,

it means for every one everyone and, to do that well,

a workforce that is a representative of the users they

served is important. They present an accelerated ap-

proach to diversity and inclusion and share, annually,

their Diversity Annual Report on how they plan to

deliver their strategy. There is a disclaimer that cur-

rent gender reporting they published is not inclusive

of their non-binary population and they intend to take

into account research such as Transgender-inclusive

measures of sex/gender for population surveys. One

of the discussed points is regarding unconscious bias,

and they work with the idea that understanding bias

and its intersection with the workplace and the com-

munities around is crucial to promote change. Sci-

ence shows that everyone is biased, once the human

brain is predisposed to negative stereotypes (Spiers

et al., 2017) and they do not expect people to rid

themselves of all bias, but we want them to recog-

nize it. Research shows that when we are more aware

of unconscious bias, we make more objective deci-

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

120

sions. To date, 84% of Google’s people managers

have taken Unconscious Bias training, and they in-

troduced Unconscious Bias workshops into all new

employees orientations. Also, Google provides nu-

merous guides to practices and tools to improve their

people processes (Google, 2018b).

Facebook (Facebook, 2018) shares, since 2014,

their journey to build a diverse company that reﬂects

the global community they serve. For 2018 report

they highlight what they believe is working and where

they can do better. Facebook believes that diversity

is critical to their success as a company once peo-

ple from all backgrounds rely on Facebook to con-

nect with others, and they will better serve their needs

with a more diverse workforce. To attract the best

and the brightest, they believe that effective recruit-

ing is critical for building a diverse company. To do

that, they work with organizations that support peo-

ple of color and women in computer science and en-

gineering, some of which include Anita Borg/Grace

Hopper (Institute, 2018), SHPE (SHPE, 2018) and

NSBE (NSBE, 2018), as well as many others that

support a broad range of groups. They see steady

increases in hiring rates for underrepresented people

since they started testing this approach in 2015.

Microsoft (Microsoft, 2018) says that diversity

and inclusion are much more than gender and race

demographics. It is about different cultures, religions,

ages, political afﬁliations, education, and sexual ori-

entations. The commitment to diversity and inclusion

means creating an environment where everyone feels

included and valued. To do that, Microsoft is commit-

ted to build and expand the pipeline for diverse tech-

nical candidates. They work with girls in early age

to sparking their interest in technology careers. Also,

work in partnerships with associations for women in

STEM and expanding their military academy pro-

gram to military bases worldwide and still to seek

meaningful ways to encourage and cultivate future

workforce. Besides, they are also reviewing their tra-

ditional recruiting practices to become more expan-

sive and more inclusive in the processes. They ex-

panded the scope of universities where they recruit,

such as Historically Black Colleges and Universities

(HBCUs) and also have programs for autism hiring.

Managers should take Inclusive Hiring training. They

believe that build a diverse culture is a critical element

to spark innovation and allow unique perspectives and

insights to the surface.

SAP (SAP, 2018) presents four diversity areas

they work on: Gender, Cross-Generational, Culture

& Identity, and Differently Abled. SAP is dedicated

to eliminating bias in the workplace and want to en-

able individuals to be recognized for what they have

to contribute. The idea is to embrace and encour-

age different perspectives and that they are stronger

by the unique combination of culture, race, ethnic-

ity, age, gender, sexual orientation, gender identity

or expression, physical or mental ability, and work-

life situations. It can be highlighted the work cross-

generational where people at different stages of life

bring a variety of perspectives and experiences to the

company — also, differently abled people and the

Autism at Work program that leverages the abilities

and perspectives of people with autism to foster inno-

vation and to help customers become intelligent enter-

prises. The program aims to reduce barriers of entry

so qualiﬁed individuals can fully develop their poten-

tial, and it employs over 140 people and in 12 coun-

tries.

Table 1: Gender Distribution in Tech Companies (Technical

Roles).

Company Female Male

Google 21.4% 78.6%

Facebook 21.6% 78.4%

Microsoft 19.0% 81.0%

SAP 33% 1] -

SAP published only the number of Women in the en-

tire company with no differentiation of kind of role.

Table 2: Ethnicity Distribution in Tech Companies.

Company Asian Black Latinx White Other

Google 41.41% 1.5% 2.8% 50.7% 3.8%

Facebook 50.3% 1.3% 3.1% 42.7% 2.6%

Microsoft 38.2% 2.7% 4.3% 52.3% 2.4%

3 CONTEXT SELECTION

StackOverﬂow

is a popular question and answer sites

for developers based on gamiﬁcation where partic-

ipants earn reputation points and badges that can

be seen as a measure of their expertise by peers

and potential recruiters and are known to motivate

users to contribute more (Vasilescu, 2014). An ex-

tensive list of academic papers using StackOverﬂow

data has been published since it was created in 2008

as Vasilescu et al.(Vasilescu et al., 2013), Bosu et

al.(Bosu et al., 2013) and, Berger et al.(Berger et al.,

2016) to mention a few. However, we did not ﬁnd

any academic paper using their Annual Developers’

Survey data (published since 2011).

Vasilescu et al. (Vasilescu et al., 2013) inves-

tigated the interplay between StackOverﬂow activi-

https://www.stackoverﬂow.com

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey

121

ties and the development process, reﬂected by code

changes committed to the social coding repository,

GitHub. Their study showed that active GitHub com-

mitters ask fewer questions and provide more answers

than others. They also observed that active Stack-

Overﬂow askers distribute their work in a less uniform

way than developers that do not ask questions. And,

ﬁnally, they showed that despite the interruptions in-

curred, the StackOverﬂow activity rate correlates with

the code changing activity in GitHub.

Once earning a high reputation score requires

technical expertise and sustained effort, Bosu et

al. (Bosu et al., 2013) analyzed the StackOverﬂow

data from four perspectives to understand the dynam-

ics of reputation building on it. The results pro-

vided guidance to new StackOverﬂow contributors

who want to earn high reputation scores quickly in-

dicating the following activities to help to build repu-

tation: answering questions related to tags with lower

expertise density, answering questions promptly, be-

ing the ﬁrst one to answer a question, being active

during off-peak hours, and contributing to diverse ar-

eas.

Berger et al. (Berger et al., 2016) studied Ques-

tion and Answer sites, like StackOverﬂow, that uses

reward systems to incentive users to answer fast and

accurately. They investigated and predicted the re-

sponse time for questions on StackOverﬂow, that ben-

eﬁt from an additional incentive through so-called

bounties. In their ﬁndings, they noted that topic re-

lated factors provide much stronger evidence than

previously found elements for these questions.

Krueger et al. (Kr

uger et al., 2017) work is about

researchers performing empirical studies in the indus-

try to gain qualitative insights into a real-world prob-

lem. However, common critics are the diversity and

selection process of participants. To address these

issues, they propose to improve the integration of

question-answering systems into an empirical study.

So, they described approaches to conduct studies in

such systems, to exemplify corresponding challenges,

and to discuss their potential. They held their research

on StackOverﬂow.

Papoutsoglou et al. (Papoutsoglou et al., 2017)

proposed a framework that aims to collect online job

advertisements from a web source which concerns In-

formation Technology job offers and to extract from

the raw text the required skills and competencies for

speciﬁc jobs. The selected professional networking

web source was StackOverﬂow, and multivariate sta-

tistical data analysis was used to test the correlations

between skills and competencies in the job offers

dataset.

Yin et al. (Yin et al., 2018) described a novel

method for extracting aligned code/natural language

pairs from StackOverﬂow. The method is based on

learning from a small number of annotated examples,

using highly informative features that capture struc-

tural aspects of the code snippet and the correspon-

dence between it and the original natural language

query.

4 METHODOLOGY

To answer our research questions, we combined data

visualization and data analysis techniques. In this sec-

tion, we present details about the dataset, the data vi-

sualization and preliminary quantitative analysis. A

more detailed discussion is done in Section 5.

4.1 Data Description

The data provided by StackOverﬂow (StackOverﬂow,

2018a) is based on a survey of 101,592 software de-

velopers from 183 countries around the world. Ac-

cordingly the criteria used by StackOverﬂow, this

number of responses are what we consider “qual-

iﬁed” for analytical purposes based on completion

and time spent on the survey; another approximately

20,000 responses were started but not included in the

analysis because respondents did not answer enough

questions. From the total qualiﬁed responses, 67,441

(66.4%) completed the entire survey. The survey was

ﬁelded from January 8 to January 28 and the median

time spent on the survey for qualiﬁed responses was

25.8 minutes, and the median time for those who ﬁn-

ished the entire survey was 29.4 minutes. Survey re-

sponses that spent less than 5 minutes were excluded

from the ﬁnal sample. Respondents were recruited

primarily through channels owned by StackOverﬂow.

Since respondents were recruited in this way, highly

engaged users on Stack Overﬂow were more likely to

notice the links for the survey and click to begin it.

Respondents who ﬁnished the survey were awarded a

“Census” badge as a motivation to complete the sur-

vey. The data is anonymized and available for down-

load in CSV format, and under the Open Database

License (ODbL). In the following section, we present

the quantitative results that answer our sub-research

questions and support the answer to the primary re-

search question of this work. First, we analyze the

aspects considered relevant by developers regarding

their interests in job opportunities. Then we see re-

spondents future goals. The third aspect we inves-

tigate is conﬁdence. We show these issues as seen

across gender, race, and ethnicity. Finally we present

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

122

Figure 1: Women/Men Priorities when Assessing a Job.

lessons learned that might serve as input for compa-

nies hiring policies.

5 RESULTS

The primary design goal for visualization is to effec-

tively communicate a thorough understanding of the

data it represents. This utility of visualization does

include usability goals but ultimately revolves around

the visualization’s ability to help people better under-

stand data (Saket et al., 2016). Nowadays many tools

are allowing to visualize and gain faster insights about

the data. In this work, we chose to use Tableau Desk-

top (Tableu, 2018) as the tool to support our visual

data analysis.

In the following sections, we present the data vi-

sualization that aims to provide support to answer the

research questions. In this work, we are not con-

sidering salary as a parameter to the discussion be-

cause it is a very well discussed subject in the in-

dustry and hard news. There are studies about gen-

der pay gaps in different areas and technology as

well (Florentine, 2018)(Tarr, )(Ismail, 2018) (Martin-

son, 2018)(Orphanides, 2018).

5.1 Aspects when Assessing Job

Opportunities

This section aims to support to answer our

RQ1: “What is being considered by developers when

assessing job opportunities?”. In the survey, Stack-

Overﬂow asked the respondents to rank ten aspects

when assessing a potential job opportunity. They

should rank from 1 (the most important) to 10 (the

least important) the following aspects:

• The industry that I’d be working in;

• The ﬁnancial performance or funding status of the

company or organization;

• The speciﬁc department or team I’d be working

on;

• The languages, frameworks, and other technolo-

gies I’d be working with;

• The compensation and beneﬁts offered;

• The ofﬁce environment or company culture;

• The opportunity to work from home/remotely;

• Opportunities for professional development;

• The diversity of the company or organization;

• How widely used or impactful the product or ser-

vice I’d be working on is.

For a better analysis of this topic, we split the an-

swers by Gender and Race/Ethnicity. The results fol-

low:

5.1.1 Gender Analysis

For women, the most important aspect when assess-

ing a job is “The ofﬁce environment or company cul-

ture”. The last important one is “The ﬁnancial perfor-

mance or funding status of the company or organiza-

tion”. For men, the distribution is quite different. Men

tend to consider ﬁrst “The compensation and beneﬁts

offered” and the last priority is “The diversity of the

company or organization”. Figure 1 compares the dis-

tribution for women and men.

When comparing the ﬁrst and last ones, the of-

ﬁce environment or company culture, that is the ﬁrst

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey

123

Figure 2: Non-Binary/Transgender Priorities when Assessing a Job.

one for women, appears as fourth for men. The ﬁ-

nancial performance or funding status of the company

or organization appears in the tenth and last place

for women and ninth place for men, showing pretty

similar importance. The compensations and beneﬁts

that appear as ﬁrst for men are in the fourth place for

women. And the diversity of the company, that ap-

pears as last important for men appears in ninth place

for women. So, even with all the discussions about

diversity individual developers that identiﬁed them-

selves as Female or Male are not making it a priority

when looking for a job.

However, the panorama changes when we evalu-

ate the data for Non-Binary and Transgender popula-

tion. Individuals that identiﬁed themselves as Non-

binaries/Genderqueer or Gender non-conforming put

ofﬁce environment or company culture and diversity

of the company as their ﬁrst and second priorities.

Transgenders put culture as the ﬁrst and diversity as

ﬁfth. The ﬁnancial performance or funding status of

the company seems to be very well aligned among all

the respondents, no matter the gender - ninth and tenth

place as shown in Figure 2.

5.1.2 Race and Ethnicity Analysis

We also analyzed the aspects when assessing a job op-

portunity by race/ethnicity. The race/ethnicity listed

in the StackOverﬂow survey were: White or of Euro-

pean descent, South Asian, Hispanic or Latino/Latina,

East Asian, Middle Eastern, Black or of African de-

scent, Native American/Paciﬁc Islander/Indigenous

Australian. Figure 3 compares assessments between

Black or Afro Descent and White or European De-

scent.

Compensation and beneﬁts are the priority

for White or European Descent, East Asians,

Hispanic/Latins and, Native American/Paciﬁc Is-

lander/Indigenous Australian. Opportunities for pro-

fessional development are priorities for Black or Afro

Descents and Middle Eastern. Languages and frame-

works were mentioned as priorities for South Asian.

However, diversity of the company is ranked in the

tenth for all. Ofﬁce environment or company cul-

ture is indicated in third by East Asians, in fourth by

White/European Descents, Native American/Paciﬁc

Islander/Indigenous Australian and Middle Eastern

and, in the ﬁfth by Hispanic/Latin and Black/Afro De-

scents.

5.2 Professional Objectives

In this section, the objective is to help to answer our

RQ2. “What are they professional objectives for the

future?”. One of the questions done in the survey

was “What Do Developers Hope To Be Doing in Five

Years?”.

5.2.1 Gender Analysis

While 26% of men want to be Working as a founder

or co-founder of their own company, near 16% of

women, non-binary and transgenders want the same.

Around 42% of women and non-binary wish to be

working in a different or more specialized technical

role in comparison to approximately 33% of men and

transgenders.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

124

Figure 3: Race/Ethnicity Priorities when Assessing a Job.

5.2.2 Race and Ethnicity Analysis

When analyzing data from the race/ethnicity point of

view, we have 44.71% Black or Afro descent wanting

to be Working as a founder or co-founder of their own

company and 41.53% of South Asians wanting to be

Working in a different or more specialized technical

role. Figure 5 shows the data for race/ethnicity

5.3 Not as Good Programming as the

Peers

This section wants to support to answer our RQ3.

“How conﬁdent developers are in their programming

skills?”.

As we can see in Figure 6, most of the answers, no

matter the gender, are on the disagreement range or in

the neither agree nor disagree. However, men tend

to disagree more with the afﬁrmation. On the scale

of agreeing and strongly agreeing, we have 29.73%

of women that agree or strongly agree that is not as

good as their peers, versus 17.28% of men that share

the same belief. 24.5% on non-binary and 19.61% of

transgenders also believe that are not as good as their

peers.

5.4 Insights from Results

In this section we intend to perform a more detailed

discussion over the data previously presented and an-

swer our more in-depth research question: “what

companies working with software engineering should

focus to attract, hire and retain talents on the specter

of diversity?”

The ﬁrst analysis presented was about the aspects

considered by developers when assessing a new job.

Diversity does not seem to be a conscious priority for

most of the developers when assessing a new job op-

portunity. However, non-binaries and transgenders

put this on the top of the rank, followed by the cul-

ture of the company. Culture appears as ﬁrst for

women and considering gender and race, always mid-

ranked. When sharing their results, StackOverﬂow it-

self mentioned: “The tech industry is struggling over-

all with issues around diversity, and individual devel-

opers are not making it a priority when looking for a

job.” (Stackoverﬂow, 2018).

It is essential to recall the case of the Google’s

employee ﬁred in 2017 after wrote a memo blaming

biology for technical’s gender gap (Varinsky, 2017).

Here we have a mix of two crucial aspects: diversity

supported by culture. Even if the diversity of the com-

pany is the last item to be considered when a devel-

oper considers a new job opportunity, companies may

consider this aspect once is linked to enterprise cul-

ture and the way to provide a safe work environment

for all.

We also evaluated what the respondents intend to

be doing in ﬁve years. Men want to be owners of

their own business when women wish to more spe-

cialized technical positions. It sounds that, devel-

opment programs that offer opportunities for women

improve their technical skills may pay off once they

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey

125

Figure 4: Gender - In Five Years.

demonstrate higher interest than men to be in special-

ized positions.

The last point analyzed is related to respondents

assessing if they believe they are not as good as their

peers on their programming skills. Women, Non-

binary and transgenders tend to doubt more their pro-

gramming skills comparing to their peers than men.

6 DISCUSSION

To perform the analysis of these results, it is essen-

tial to recall previous events. In 2016, Google pub-

lished a study about the diversity gap in computer sci-

ence (Inc. and Inc., 2016). In this study, they iden-

tiﬁed that male student are more interested and more

conﬁdent in learning computer science, and that fe-

male students rate themselves lower in skills related

to Computer Science. Another point identiﬁed by the

study is that stereotypes may inﬂuence implicit be-

liefs about who can study computer science and might

introduce unconscious bias in educators and parents,

who may disproportionately and unconsciously en-

courage students who ﬁt the computer scientist stereo-

type to pursue Computer Science. For example, male

students are more likely than female students to have

been told by a teacher (39% vs. 26%) or a parent

(46% vs. 27%) that they would be good at Computer

Science. Teachers and parents may reinforce stereo-

types by telling more male students they think they

would be good at Computer Science, thus furthering

the underrepresentation of females in Computer Sci-

ence.

Recent research published in Nature from O’Dea

et al. (O’Dea et al., 2018) says that girls are suscep-

tible to conforming to stereotypes in the traditionally

male-dominated ﬁelds of STEM, and backlash effects

hinder girls who try to succeed in these ﬁelds. A

girl’s answer to the question of “what do you want

to be when you grow up?” will be shaped by her

own beliefs about gender, and the collective beliefs

of the society she is raised in. The study also com-

pares school grades of girls and boys. They found

out that girls tend to earn higher school grades than

boys, including in STEM subjects. The prediction of

the grade distribution represents that, when all grades

are considered, girls on average earn higher grades

and are less variable than boys, although there are

more highly performing boys than girls at the upper

end of the achievement distribution. Therefore, by the

time a girl graduates, she is just as likely as a boy to

have earned high enough grades to pursue a career in

STEM. When she evaluates her options, however, the

STEM path is trodden by more male competitors than

non-STEM and presents additional internal and exter-

nal threats due to her and societies’ gendered beliefs

(stereotype threat and backlash effects). Additionally,

the paper says that gender differences in expectations

of success can arise due to backlash effects against

individuals who defy the stereotype of their gender,

or due to gender differences in ‘abilities tilt’ (having

comparatively high ability in one discipline compared

to another). Women in male-dominated pursuits, in-

cluding STEM, face a paradox: if they conform to

gender stereotypes, they might be perceived as less

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

126

Figure 5: Race/Ethnicity - In Five Years.

competent, but if they defy gender stereotypes and

perform ‘like a man,’ then their progress can be halted

by ‘backlash’ from both men and women.

So, the results observed in the StackOverﬂow De-

velopers’ Survey may be a reﬂex of stereotypes built

from previous personal experiences and be directly re-

lated to a conﬁdence gap in women, mainly.

An important point to be considered is what

is called the impostor syndrome. Jackson and

Heath (Jackson and Heath, 2014) says that impostor

syndrome is deﬁned as a psychological phenomenon

in which people are unable to internalize their accom-

plishments. Impostor syndrome affects most people

at some point during their careers across all races, all

genders, and all ages.

Sukhai (Sukhai and Mohler, 2016) mentions that

impostor syndrome is common within the academic

environment, particularly at the graduate level stu-

dent - in STEM ﬁelds, mainly, where productivity is

a signiﬁcant measure of a student’s success. Impos-

tor syndrome presents itself as a series of feeling or

thoughts, and one of them is the frustration with the

inability to meet self-set standards (“I will never be

as good as I want to be, so why bother trying?”).

Churchill (Churchill, 2018) points out that over four

decades ago when it was given a name by clinical psy-

chologists Pauline Clance and Suzanne Imes in the

late 1970s, this feeling was prevalent among high-

achieving women.

Follow-on research has shown that impostor syn-

drome is very real and very prevalent and that its

effects are undeniably negative. Therefore also un-

surprising that there is a strong correlation between

impostor syndrome and anxiety, stress, depression,

and burnout, the debilitating condition of exhaustion

that can result in talented individuals giving up on

promising careers. StackOverﬂow shares the results

about disability status in their results page, with num-

bers about anxiety, depression and focus (StackOver-

ﬂow, 2018b), but, unfortunately, the individualized

and anonymized results about it were not shared in

the CSV ﬁle, and we could not make the gender cor-

relation.

To overcome it, it is essential to have a support

network that helps to identiﬁed impostor syndrome in

their workforce. There are some strategies to over-

come the syndrome, for example: instruct employ-

ees that comparison with others must be done with

care (Jackson and Heath, 2014). Comparison with-

out context can be misleading (people will compare

themselves with people in another skill level).

That said, what we found in the data from of

StackOverﬂow Developers’ Survey suggests the im-

portance of the initiatives to minimize bias and stereo-

types that some companies are doing in their hiring

process and the process of development of their tech-

nical team.

7 THREATS TO VALIDITY

The validity of this work can be subjected to some

threats. In the following, threats to internal validity

and external validity are illustrated.

External validity refers to how much we can gen-

eralize our ﬁndings. The presented results are based

on data from the StackOverﬂow community. We sus-

pect that given the high number of respondents and

the reputation of the community the results could be

generalized outside the scope of our study.

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey

127

Figure 6: Not as good programming as the peers.

Internal validity often refers to experimenter bi-

ases. For our results, the threat of misreading the

data visualization analysis and getting to conclusions

based on the knowledge areas of the researchers in-

volved in this work.

8 CONCLUSION

The present work set out to provide insights to sup-

port the attraction, hiring and retention policies for

more diverse and inclusive software engineering en-

vironments. Using the anonymized data from Stack-

Overﬂow Developer’s Survey, we performed analysis

and correlations beyond their original ones with the

support of data visualization techniques that implied

in insights to our recommendations. Results show

that diversity in the company is not yet a full con-

scious decision-making factor for developers assess-

ing a new job opportunity, and respondents that iden-

tiﬁed themselves as women, non-binary and transgen-

ders tend to doubt more their programming skills be-

lieving they are not as good as their peers. A dis-

cussion about the unconscious bias, stereotypes, and

impostor syndrome was done, and we reinforce the

importance of initiatives to minimize bias and stereo-

types that companies are doing in their hiring pro-

cess and the process of development of their technical

team.

For future work studies, we see opportunities

when we selected more speciﬁc aspects in the spec-

trum of the diversity. For example, for cognitive di-

versity, since there has been an increase in computer

science students with the Asperger Syndrom (Ribu,

2010)(Egan, 2005), it is also important to tackle this

issue globally in Software Engineering. There is also

a need for teaching institutions and software compa-

nies to work together to understand these differences

better to include them.

ACKNOWLEDGEMENTS

This project is partially funded by FAPERGS, project

17/2551-0001/205-4.

REFERENCES

Atadero, R. A., Paguyo, C. H., Rambo-Hernandez, K. E.,

and Henderson, H. L. (2018). Building inclusive en-

gineering identities: implications for changing engi-

neering culture. European Journal of Engineering Ed-

ucation, 43(3):378–398.

Berger, P., Hennig, P., Bocklisch, T., Herold, T., and Meinel,

C. (2016). A journey of bounty hunters: Analyzing the

inﬂuence of reward systems on stackoverﬂow ques-

tion response times. In 2016 IEEE/WIC/ACM Inter-

national Conference on Web Intelligence (WI), pages

644–649.

Bosu, A., Corley, C. S., Heaton, D., Chatterji, D., Carver,

J. C., and Kraft, N. A. (2013). Building reputation

in stackoverﬂow: An empirical investigation. In Pro-

ceedings of the 10th Working Conference on Mining

Software Repositories, MSR ’13, pages 89–92, Pis-

cataway, NJ, USA. IEEE Press.

Churchill, E. F. (2018). Is there a ﬁx for impostor syn-

drome? Interactions, 25(3):22–24.

Egan, M. A. L. (2005). Students with asperger’s syndrome

in the cs classroom. In ACM SIGCSE Bulletin, vol-

ume 37, pages 27–30. ACM.

Facebook (2018). Facebook diversity update.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

128

Florentine, S. (2018). The hidden truths of it’s gender pay

gap.

orster, A. (2018). Ethnic heterogeneity and electoral

turnout: Evidence from linking neighbourhood data

with individual voter data. Electoral Studies, 53:57–

65.

Google (2018a). Google diversity.

Google (2018b). Rework guides.

Inc., G. and Inc., G. (2016). Diversity gaps in computer

science: Exploring the underrepresentation of girls,

blacks and hispanics.

Institute, A. B. (2018). Anita borg/grace hopper.

Ismail, N. (2018). Is there a gender pay gap in the technol-

ogy sector?

Jackson, D. and Heath, T. (2014). An antidote to impostor

syndrome. XRDS, 21(2):12–13.

uger, J., Schr

oter, I., Kenner, A., and Leich, T. (2017).

Empirical studies in question-answering systems: A

discussion. In Proceedings of the 5th International

Workshop on Conducting Empirical Studies in Indus-

try, CESI ’17, pages 23–26, Piscataway, NJ, USA.

IEEE Press.

Martinson, J. (2018). Technology will widen pay gap and

hit women hardest – davos report.

Menard, P., Warkentin, M., and Lowry, P. B. (2018). The

impact of collectivism and psychological ownership

on protection motivation: A cross-cultural examina-

tion. Computers and Security, 75:147–166.

Microsoft (2018). Microsoft global diversity and inclusion.

Mike Beedle, Arie van Bennekum, A. C. W. C. M. F. J. H.

A. H. R. J. J. K. B. M. R. C. M. K. S. J. S. D. T. (2001).

Agile manifesto.

NSBE (2018). National society of black engineers.

O’Dea, R. E., Lagisz, M., Jennions, M. D., and Nakagawa,

S. (2018). Gender differences in individual variation

in academic grades fail to ﬁt expected patterns for

STEM. Nature Communications, 9(1):3777.

Orphanides, K. (2018). Gender pay gap: how big tech com-

panies in the uk compare.

Page, S. E. (2007). The Difference: How the Power of

Diversity Creates Better Groups, Firms, Schools, and

Societies. Princeton University Press.

Page, S. E. (2017). The Diversity Bonus: How Great Teams

Pay Off in the Knowledge Economy (Our Compelling

Interests). Princeton University Press.

Papoutsoglou, M., Mittas, N., and Angelis, L. (2017).

Mining people analytics from stackoverﬂow job ad-

vertisements. In 2017 43rd Euromicro Conference

on Software Engineering and Advanced Applications

(SEAA), pages 108–115.

Pressman, R. (2010). Software Engineering: A Practi-

tioner’s Approach. McGraw-Hill, Inc., New York,

NY, USA, 7 edition.

Ribu, K. (2010). Teaching computer science to students

with asperger’s syndrome. E. Hjelm

as (red.) Norsk

Informatikkonferanse, pages 99–111.

Saket, B., Endert, A., and Stasko, J. (2016). Beyond us-

ability and performance: A review of user experience-

focused evaluations in visualization. In Proceedings

of the Sixth Workshop on Beyond Time and Errors on

Novel Evaluation Methods for Visualization, BELIV

’16, pages 133–142, New York, NY, USA. ACM.

SAP (2018). Sap diversity.

SHPE (2018). Society of hispanic professional engineers.

Spiers, H. J., Love, B. C., Le Pelley, M. E., Gibb, C. E.,

and Murphy, R. A. (2017). Anterior temporal lobe

tracks the formation of prejudice. Journal of Cognitive

Neuroscience, 29(3):530–544. PMID: 27800703.

Stackoverﬂow (2018). How do developers assess potential

jobs?

StackOverﬂow (2018a). Stackoverﬂow developers survey

2018.

StackOverﬂow (2018b). Stackoverﬂow survey - developer

proﬁle disability status.

Sukhai, M. and Mohler, C. (2016). Creating a Culture of

Accessibility in the Sciences. Academic Press.

Tableu (2018). Tableau.

Tarr, T. By the numbers: What pay inequality looks like for

women in tech.

Varinsky, D. (2017). Google - employee ﬁred.

Vasilescu, B. (2014). Software developers are humans,

too! In Proceedings of the Companion Publication

of the 17th ACM Conference on Computer Supported

Cooperative Work & Social Computing, CSCW

Companion ’14, pages 97–100, New York, NY, USA.

ACM.

Vasilescu, B., Capiluppi, A., and Serebrenik, A. (2014).

Gender, representation and online participation: A

quantitative study. Interacting with Computers,

26(5):488–511.

Vasilescu, B., Filkov, V., and Serebrenik, A. (2013). Stack-

overﬂow and github: Associations between software

development and crowdsourced knowledge. In 2013

International Conference on Social Computing, pages

188–195.

Yin, P., Deng, B., Chen, E., Vasilescu, B., and Neubig, G.

(2018). Learning to mine aligned code and natural

language pairs from stack overﬂow. In Proceedings of

the 15th International Conference on Mining Software

Repositories, MSR ’18, pages 476–486, New York,

NY, USA. ACM.

Reinforcing Diversity Company Policies: Insights from StackOverﬂow Developers Survey

129