PreprintPDF Available

Understanding Equity, Diversity and Inclusion Challenges Within the Research Software Community

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

Research software -- specialist software used to support or undertake research -- is of huge importance to researchers. It contributes to significant advances in the wider world and requires collaboration between people with diverse skills and backgrounds. Analysis of recent survey data provides evidence for a lack of diversity in the Research Software Engineer community. We identify interventions which could address challenges in the wider research software community and highlight areas where the community is becoming more diverse. There are also lessons that are applicable, more generally, to the field of software development around recruitment from other disciplines and the importance of welcoming communities.
Understanding Equity, Diversity and Inclusion
Challenges Within the Research Software
Community
Neil P. Chue Hong1[0000000288767606], Jeremy Cohen2[0000000343122537],
and Caroline Jay3[0000000260801382]
1Software Sustainability Institute & EPCC, University of Edinburgh, Edinburgh,
U.K.
N.ChueHong@epcc.ed.ac.uk
2Department of Computing, Imperial College London, London, U.K.
jeremy.cohen@imperial.ac.uk
3Department of Computer Science, University of Manchester, Manchester, U.K.
Caroline.Jay@manchester.ac.uk
Abstract. Research software – specialist software used to support or
undertake research – is of huge importance to researchers. It contributes
to significant advances in the wider world and requires collaboration
between people with diverse skills and backgrounds. Analysis of recent
survey data provides evidence for a lack of diversity in the Research Soft-
ware Engineer community. We identify interventions which could address
challenges in the wider research software community and highlight areas
where the community is becoming more diverse. There are also lessons
that are applicable, more generally, to the field of software development
around recruitment from other disciplines and the importance of wel-
coming communities.
Keywords: research software, software engineering, research software
engineering, diversity, EDI.
1 Introduction
Developing specialist research software to support computational science is an es-
pecially challenging process. Unlike more traditional software engineering tasks,
researchers or, increasingly, RSEs (Research Software Engineers) who write this
software need an understanding of the underlying scientific challenge being ad-
dressed.
The methods used in research favour a hypothesis-driven approach. This
creates a different working environment from the wider software engineering
industry, where software is built to meet a client’s specification. As noted by
Hettrick et al.,
To be effective, software development in research should be approached,
not as a one-off transaction, but as a partnership between researcher and
software expert.” [3]
arXiv:2104.01712v1 [cs.SE] 4 Apr 2021
2 N.P. Chue Hong et al.
The “partnership between researcher and software expert” mentioned here
highlights a need for varied skills and good communication. It is clear that much
computational research requires diversity of skills and experience, working in
partnership, that spans software engineering and science. Researchers are more
likely to be working with others who have different technical expertise, use dif-
ferent technical terminology, and may be communicating in a tertiary language.
Diversity in project teams, and workplaces in general, is important and it
is widely accepted that there are many benefits from ensuring diversity within
teams and communities. For example, there is evidence to show that diversity
in terms of knowledge or skills can be beneficial [19] and that gender diversity
within groups can result in higher quality scientific outputs [4]. However, there
are cases where diversity can raise challenges, for example with reduced feelings
of well-being amongst members in highly-diverse teams [31]. Diversity in the
context of teams in industry and research is an area that has been the subject
of extensive research and there are some contradictory results from the many
studies undertaken. Stahl et al. [26] look at a number of different studies of cul-
tural diversity concluding that there are both benefits and drawbacks. Whether
benefits can be realised while minimising more challenging aspects will depend
on effective process management. In the context of software engineering, Capretz
& Ahmed [5] looked at how different personality traits relate to suitability for
different roles in software projects. Ultimately they conclude that diversity in
terms of personalities and skills is important in helping with problem solving
tasks involved in building and maintaining software.
The different types of diversity highlighted so far, including diversity of skills
and knowledge, culture, personality and gender are, of course, just some of the
many different aspects that lead to diversity amongst a group of individuals.
Others, among a huge range, include ethnicity, disability and age.
Research Software Engineers are, at present, much more likely to come from
a research background than software engineering professionals working in other
fields. Nonetheless, we know from surveys within the RSE community that a
similar diversity crisis to that identified in other fields exists. In this paper,
we examine current problems with diversity in research software engineering
and consider potential causes. Our aim is to present a better understanding
of existing diversity within the RSE community in order to develop insights
and recommendations to address current issues. We achieve this through an
empirical analysis of existing open data collected and made available through
large-scale international surveys of research software engineers undertaken by the
RSE community. We also review related work from allied areas which suggests
ways of addressing the challenges identified.
As pointed out by Mathieson in [20], based on 2018 ONS data, just one in
eight of almost 340,000 software development professionals in the UK are women.
This example demonstrates that there is a long way to go to address the lack of
diversity in the wider software engineering domain, as well as the more focused
domain of research software engineers. Professional software engineers can gain
their skills through a variety of different routes and they may come from a wide
Understanding EDI Challenges Within the Research Software Community 3
range of different disciplines. For example, they may have developed their skills
through a degree programme or vocational training in computing or a related
area. Alternatively, they may come from a completely different disciplinary back-
ground and have re-trained through one of a large number of “coding schools”
that offer training, often via intensive courses, in software development skills.
The wide array of routes into the field creates an expectation of a diverse field
which makes it all the more surprising that this is not the case.
We highlight three core contributions this paper provides to understanding
the importance of diversity among RSEs working in computational and data
science:
Providing an evidence-based analysis that demonstrates the current prob-
lems with diversity in the research software engineering community.
Demonstrating that there is already extensive “domain mobility” for RSEs
and that a lack of such mobility is therefore not likely to be a cause for a
lack of diversity within the field.
Offering four general recommendations that we believe can form a basis to
support addressing the lack of diversity in research software engineering.
In Section 2we examine the International RSE Survey results from 2018
and undertake further analysis on this data. Section 3highlights the three areas
that we see as both helping to explain and provide the basis for addressing the
lack of diversity amongst RSEs, while discussion and conclusions are provided
in Section 4.
2 Surveying RSE Diversity
To understand the diversity challenges facing those embarking on careers as
Research Software Engineers, we require a better understanding of the landscape
as a whole. The Software Sustainability Institute coordinates an ongoing series
of surveys of the RSE community. In the most recent survey [24], from 2018,
participants were asked a number of socio-demographic questions.
Percentage of UK
who are: RSEs[24] Academics[15]Software
Developers[25]
All UK
workers[25]
Gender (female) 14 46 14 48
Ethnicity
(BAME/Mixed) 5 15 21 12
Report disability 6 4 10 13
Table 1. Comparison of 2018 demographics for Research Software Engineers, Aca-
demics, Software Developers and general working population in the United Kingdom.
Our reanalysis primarily focuses on UK data, as this is where the authors
are located, but the numbers from other countries in the survey are broadly
similar. Table 1compares RSEs with UK Higher Education Statistics Authority
(HESA) data [15], and a study by the British Computer Society (BCS) of soft-
ware development professionals in the UK [25] based on 2018 Office for National
4 N.P. Chue Hong et al.
Statistics data. Limitations in the RSE data currently available to us mean that
we focus on aspects of gender, ethnicity and disability, as opposed to other types
of diversity.
This comparison indicates a gender diversity gap, which we might expect
given the percentage of women working as software developers is also 14%. But
this could be better: software is a fundamental part of all research, and 46% of
academic staff are female. The ethnicity data indicates a greater problem. Of
the respondents who declared their ethnicity, there were only 6 non-white and
5 mixed race RSEs – 5%. These figures are significantly lower than the 28%
BAME students studying computer science [30] and 10% of BAME physicists
[17]. Although we should not aggregate non-white ethnic categories together, as
they will experience different challenges and biases, it is striking that RSEs do
not fit the general profile of those working in the IT industry, where there is
greater ethnic diversity.
Fig. 1. Flow of UK RSEs from academic discipline studied to domains they now work
in, categorised by gender: 79.8% male (green) / 14.3% female (purple) / 5.9% other
responses (grey). N=203.
Understanding EDI Challenges Within the Research Software Community 5
An important area of research is understanding if diversity can be improved if
RSEs are drawn from a wider set of degree backgrounds. Figure 1shows where
RSEs are working based on their academic degree. Only gender diversity was
considered, given the low ethnic diversity and number of respondents reporting
a disability. Gender balance is relatively uniform across fields rather than mir-
roring the gender balance in that field. Over half of RSEs have a first degree
in Physics and Astronomy or Computer Science. In the UK, 17% of CS under-
graduates were female compared with 41% of physical sciences undergraduates
[16], suggesting that RSEs come from the “computational” subset of a subject.
However, within computer science research within the UK just under 23% of
academics and researchers are female [14]. This perhaps suggests that a larger
percentage of female CS undergraduates move on to a research or faculty posi-
tion in CS than male undergraduates, or that female CS researchers are moving
into the field after undergraduate studies in other areas. However, it could also
be attributed to other factors such as industry hiring trends. Of interest here
is that, while over half of RSEs have a Physics and Astronomy or Computer
Science-focused first degree, the percentage of female RSEs is still somewhat
below the percentage of female CS research/faculty staff.
The areas that RSEs support are more widespread than the fields they are
currently recruited from. There are teams offering general RSE support at a
number of institutions and many RSEs working in the biosciences, geosciences,
and medicine. This suggests that a challenge research software engineering has
is encouraging more people with first degrees in these areas (which have more
balanced gender representation) to become RSEs. This is important because
domain knowledge can be especially valuable when undertaking RSE projects.
Things to consider include the wording of job adverts, which can affect who ap-
plies for them [1] — can we encourage more candidates from other disciplines?
There may be lessons for the software development field more generally — there
are a significant number of RSEs whose background is in biological sciences or
geosciences, as well as some from social sciences and the humanities, showing
people with “non-traditional” degree backgrounds seek careers as software prac-
titioners.
One encouraging aspect of Figure 1is there appear to be many lines going
from one discipline on the left hand side to a different discipline on the right
hand side, representing an RSE working in an area different from the one where
they trained. This suggests that there is scope to improve diversity within the
RSE community through attracting individuals from more diverse domains into
the RSE space. This provides an opportunity for more immediate improvements
in the diversity of the RSE community than the multi-year timeline that we
might expect for efforts to improve diversity amongst undergraduate cohorts. To
investigate this movement between domains, which we call “domain mobility”,
we undertook further analysis of the most recent 2018 RSE survey data.
Table 2shows an example of “domain mobility” for a series of domains where
the number of survey respondents was >20. We define partial mobility as RSEs
working in one or more domains outside the domain of their highest degree, while
6 N.P. Chue Hong et al.
Discipline N Partially
mobile (%)
Fully mobile
(%)
Biological Sciences 95 44.21 8.42
Chemistry 22 40.91 22.73
Computer Science 241 51.45 22.41
Electrical & Electronic Engineering 36 41.67 38.89
Geography & Environmental Sciences 46 52.17 6.52
Mathematics 69 47.83 47.83
Physics and Astronomy 261 31.80 19.92
Table 2. Example of domain mobility across a selection of disciplines from the 2018
RSE survey data where N>20.
also still undertaking work in the domain of their degree. Full mobility applies a
further filter to partial mobility by excluding RSEs who still work in the domain
of their highest degree. So, in the case of Mathematics, where the figures for
partial and full mobility are the same, none of the survey respondents who said
their highest degree was in Mathematics undertake RSE work in the domain
of Mathematics. It should be noted that while domain mobility is a positive
concept, and the ability to work across a wide variety of different domains is
a great benefit, we do not have any data on the reasons for these movements
between domains. The movements may be individuals wanting to undertake
RSE work in, and learn about, a different domain to their original area of study.
However, there are also likely to be cases of movement between domains being
made out of necessity due to a lack of job opportunities in an individual’s domain
of choice, for example.
2.1 Code and Data Availability
The Jupyter Notebooks used to perform the analysis are available from [8].
International RSE data can be obtained from [24], apart from some gender data
which have not been publicly published at the time of writing.
3 Improving EDI in the research software community
In this section we investigate three areas that represent both challenges and
opportunities for improving equity, diversity and inclusion (EDI) within the re-
search software community. Through the material in this section, we hope to
highlight possible reasons for the lack of diversity in the research software space
while also providing some thoughts and guidance on how these can be addressed
and how the community can work together to help improve a wide range of
aspects of diversity. The material here is aimed at both developers of research
software, and at team leaders, technical managers and individuals in other man-
agement and community leadership roles. For individuals who write software,
we aim to offer thoughts and advice that can help improve understanding of
the benefits of diversity and recognise situations that can contribute to a lack
of diversity. For managers and team leaders, we hope to improve recognition of
opportunities to address diversity challenges, both locally within teams and in
the wider community in the context of events and activities.
Understanding EDI Challenges Within the Research Software Community 7
3.1 Safety in similarity?
Diversity, whether in terms of gender, ethnicity, skills or other characteristics, is
important in bringing different perspectives, ideas and experiences to a commu-
nity. Diverse teams can lead to higher quality science [4] and improved technology
business performance [22].
However, as highlighted by Merritt [21] in the context of hiring staff, indi-
viduals want to work with people who are like them and Lang & Liu [19] point
out contradictory evidence for the benefits of diversity within teams. They high-
light work by Byrne in the 1971 book “The Attraction Paradigm” that suggests
certain similarities within teams can be helpful in ensuring effective working en-
vironments. However, they do also highlight a series of other work that supports
the idea of diversity within teams and groups being beneficial. Van der Zee, et
al. [31] highlight previous work that suggests that we tend to have a positive re-
sponse to similarity while the opposite is true for dissimilarity. They also point
out other previous work suggesting that we have an attraction to people who
share similar attitudes and values to our own since this makes communication
with them easier.
A very large study of over 20 years of scientific papers [12] shows that paper
co-authors are more likely to be of similar ethnicity. However, where papers are
the result of collaborations between authors at different locations, it was shown
that this can result in them being published in journals with high impact factors
and receiving more citations.
The above examples demonstrate that there is a perceived “safety” in being
around people who have similar interests, backgrounds and/or values. We see
that this is likely to work both ways and that in the case of a community,
such as the RSE community or groups of software practitioners more generally,
individuals choose whether or not to engage partly based on whether they see
people like themselves within that community. Of course, while individuals may
seek out others who they feel are similar or who they have things in common
with, diversity can be extremely important and hugely valuable in providing
different views, ideas, attitudes and perspectives. Ultimately this can lead to
important benefits, even if there may be learning experiences to be had along
the way. These include, for example, getting more used to working with people
who may not look at every opportunity, challenge or research problem in the
same way. Many of us will have had experiences that support this when attending
community events and workshops which often involve breakout group discussions
or problem-solving tasks.
Awareness is an important step here and this begins with educating team
leaders, community managers and event organisers to be more aware
of situations where there is a lack of diversity and the benefits of ad-
dressing this. Developing a stronger understanding of the benefits of diversity,
but also the challenges that can arise in ensuring that diverse groups can collab-
orate effectively, is an important step towards addressing the lack of diversity
in the scientific software community. This is especially important for individuals
in leadership or organisational roles. Various guides, for example the Hopper
8 N.P. Chue Hong et al.
Conference Diversity Guide [28] and NumFOCUS DISCOVER Cookbook [23],
provide detailed information on approaches for helping to ensure diversity at
conferences and events. There is also a role for RSE group leaders in encourag-
ing and enabling the RSEs in their teams to collaborate with other institutions,
as this diversity might lead to higher impact of their work.
3.2 Increasing equity, diversity and inclusion at events
The events that we consider in this section include everything from large confer-
ences and workshops to small local community events with only a small number
of attendees. Large events often have dedicated organising teams and poten-
tially co-chairs specifically dedicated to areas such as EDI. This is unlikely to
be the case for a small community event with perhaps only a few 10s of par-
ticipants. Nonetheless, being able to widen participation at events requires you
to start with an understanding of the community the event is targeted at and
an awareness of the particular aspects of diversity that you aim to improve [9].
One suggestion for improving equity and inclusion at conferences, from work
looking at 30 conferences in the conservation and ecology domain, is to write up
and promote details of the actions and processes followed to support improving
EDI [29]. This work demonstrates that concerns around ensuring diversity at
conferences and events are not specific to the research software community. The
sort of open approach espoused through supporting and promoting EDI activi-
ties could be beneficial across many, if not all, domains. It would help to offer
a demonstration of a conference’s efforts to support aspects of EDI and also
provide evidence for the wider community to help identify what works and what
is less successful. This should help to avoid repeating less successful approaches
to addressing diversity and inclusion concerns across different domains.
Ensuring that event committees, speakers and panel members reflect the di-
versity that organisers would like to see among an event’s attendees has been
effective in some fields. This builds on some of the ideas highlighted in Sec-
tion 3.1. Research that looked at presenters across 21 meetings of the American
Association of Physical Anthropologists found events that had either female, or
both female and male organisers, resulted in a much higher percentage of women
as first authors of presented papers and posters [18]. However a study by Bano
and Zowghi of six software engineering conference series [2] found that having a
female conference chair or program committee chair did not significantly affect
the number of keynote speakers that were female, or the make up of the program
committee. This work did not, however, look at all conference speakers.
While the figures shown in Section 2show a lack of diversity within the RSE
community, there is a perception, when attending RSE workshops and events, of
gender and ethnic diversity being significantly better than the headline survey
figures suggest. RSE events generally have a significant number of female speak-
ers, workshop organisers and community leaders. We see with the Collaborations
Workshop series, an event that focuses on general research software practice, a
positive change in the diversity balance. It has had a policy of ensuring diver-
sity of keynote speakers, and in 2020, 55% of steering committee members were
Understanding EDI Challenges Within the Research Software Community 9
female, 42% of all speakers were female, 34% of attendees reported their gender
as female, and 20% of attendees reported their ethnicity as non-white or mixed.
Nonetheless, the challenge remains of attracting more individuals from a
wider range of backgrounds to get involved with RSE. A concrete action that
can be taken here is to work to increase the diversity amongst organisers,
speakers and sponsors at RSE events and be more open about the
approaches taken to support this. The latter part of this recommendation
should help organisers of other events, both within the research software commu-
nity and beyond, to learn from and build on efforts to improve EDI. Ultimately
this should help to accelerate the process of improving diversity throughout the
research community.
3.3 An inclusive culture and safe space
An inclusive culture is an important foundation of diversity [6]. Research teams
that include RSEs frequently have only one or two individuals in RSE roles.
Having valuable but different skills, and different career aims, can make RSEs
feel like the “odd one out” within a research group or team. As a result, RSE
communities of practice have developed in recent years to support individuals
in this field. Possibly due to previous perceived marginalisation, they tend to be
open, welcoming communities. These communities provide a great opportunity
to meet, network and collaborate with others who understand the challenges
RSEs face in their day-to-day work.
At conferences and events, one way of formally promoting inclusion is hav-
ing a clear and well-publicised Code of Conduct. The importance of a Code of
Conduct and the challenges it can help to address and provide guidance on are
highlighted in [10]. While still not commonplace at traditional academic con-
ferences [11], it is unusual to find an RSE conference or workshop that does
not have a Code of Conduct – learning from the experience of the open source
community. To ensure best practice and clarity of the message provided, many
events choose to build on widely accepted, open codes of conduct such as the
template provided by the Geek Feminism wiki [13], and also have a diversity
statement.
In the context of workplaces and research teams, it is also important that
individuals feel that they fit in, both within the workplace environment itself and
with the people based there. Cheryan et al. [7] call this “ambient belonging”.
They undertook a series of studies within a computer science context looking
at gender-based perspectives on the influence of different environments. This
work showed that environments that had aspects that made them fit with com-
puter science stereotypes reduced the interest of women participating in the
field. Changing the environments to make them appear less like something that
would be associated with computer science increased interest in the field. This
provides an example of the sort of potentially small, but nonetheless significant,
changes that can help to develop an inclusive culture, a welcoming environment
and, ultimately, help to improve diversity. We consider that there may also be
opportunities to help improve other aspects of diversity through identifying and
10 N.P. Chue Hong et al.
making similarly small environmental or organisational changes in different con-
texts, for example to improve the experience for people with disabilities.
The wide ranging use of codes of conduct at research software events and
the recognition of the importance of communities in helping to provide individ-
uals with a place where they fit in is significant. It suggests that the research
software community is developing events that are explicit in their desire to be
open and diverse, with clear statements on acceptable behaviour. We see that
progress is being made in developing an inclusive culture and providing envi-
ronments through events and activities that potential participants feel are open
and welcoming. It is of vital importance that this continues. There are two
concrete recommendations from this analysis. Firstly, within RSE groups,
both team leaders and members should be aware of the importance
of their working environment and how potentially small environmen-
tal changes may affect other team members’ feelings of inclusion and
belonging. Secondly, in the context of community events and activities, event
organisers should look to highlight their support for diversity and in-
clusion through the use of a diversity statement and provide a clear
code of conduct highlighting acceptable behaviour.
The Software Sustainability Institute runs a fellowship programme recognis-
ing the diverse roles and skills of those working to promote research software
practice. An evaluation of the programme showed it plays an important role in
supporting communities of best practice and skills transfer, and that a signifi-
cant benefit is the way it has raised the profile of software in research, and those
people who develop and advocate for it [27]. This has had positive effects for
those who may previously have considered themselves as ‘outsiders’ in the role,
or lacked confidence. This is exemplified by a comment from a female respondent:
Despite getting a PhD partially from a computer science programme, I
could see that my skills and knowledge were always at least to some extent
dismissed or doubted [...] since being elected a SSI fellow I most definitely
observed a significant drop in mansplaining... I have little doubt that the
SSI fellowship was a significant [reason] I got my current position as
(Head of Division at a Supercomputing Center).” [27]
4 Discussion and Conclusions
Our research has shown that the gender diversity of RSEs is similar to the field
of software engineering, but does not reflect the gender balance of the academic
research workplace. This is particularly true where RSEs are working in domains
such as the biosciences, geosciences and medicine. Ethnic diversity among RSEs
is worse than in the wider software industry, but it is unclear why. The num-
ber of RSEs with disabilities reflects the academic research workplace, and is
poor in comparison to the IT profession. Given that 19% of the UK working age
population report a disability, there is much to be done to provide a equitable
workplace for RSEs with disabilities. Differences in workplace culture, environ-
ment or incentives may be factors, and further research is required. However,
Understanding EDI Challenges Within the Research Software Community 11
general interventions to improve diversity appear to be increasing gender and
ethnic diversity at events.
We have highlighted the importance of communication and collaboration be-
tween individuals building software to support research and the computational
scientists and researchers that they collaborate with. We have also shown that
diversity among all parties in these collaborations can lead to better communi-
cation, a wider range of ideas and perspectives, and, ultimately more effective
collaborations that produce higher quality outputs. The three areas discussed
in Section 3as key opportunities for improving equity, diversity and inclusion
within the field of research software can be summarised as follows:
– Safety in similarity? The perceived safety of looking to collaborate and
work with individuals who have similar interests, backgrounds or values is
widely recognised. While there is research that suggests there can be bene-
fits to this approach, there is also extensive work highlighting the benefits
of diversity. However, this is something that can be approached from two
different perspectives and individuals may be more likely to join and engage
with a community if they see people like themselves within that community.
Increasing diversity at events: It is important to highlight and promote
actions being take to improve EDI in the context of events. It is also impor-
tant to ensure that both organising groups and speakers reflect the level of
diversity that an event’s organisers would like to see among its participants.
An inclusive culture and safe space: This is a very important aspect of
both participating in a community, and being part of an RSE group, for ev-
eryone involved. A well-publicised Code of Conduct and diversity statement
with clearly defined processes supporting them are key elements in helping
to ensure this within the context of events. In RSE groups, leaders and team
members need to be aware of the importance of an inclusive culture and
working environment.
Poor diversity in the community of research software developers is likely a
result of the low levels of diversity in the disciplines that currently form the ma-
jor path towards an RSE role or career. We have identified scope to widen the
range of areas from which RSEs are recruited and it is hoped that this can be
achieved by better use of language in the way that RSE roles are advertised and
in the way that RSE is promoted, more generally. This may also have relevance
for professional software practitioners looking to get involved in the research
community, as it is clear that there are many people with “non-traditional” de-
grees seeking careers developing software. There is a role for professional bodies,
such as the British Computer Society, Association for Computing Machinery and
IEEE Computer Society, to support and embrace these career paths and help
improve the gender diversity at undergraduate and postgraduate level.
Nonetheless, we feel that despite the knowledge gained from our extended
analysis of existing survey data, and the pre-existing material that we have
referenced, this is a hugely complex but very important area. It would benefit
extensively from additional evidence that could be gathered through a range
12 N.P. Chue Hong et al.
of further empirical studies. Some of the areas that may be considered most
important for such studies in the short term include:
Exploring why levels of diversity among RSEs are lower than in many of
the areas of study and research that already feed into RSE careers, such
as Computer Science and Physics and Astronomy. What are the levels of
diversity in other career paths for these subjects? What influences the career
choices that individuals make and are there specific aspects that steer them
away from an RSE career?
Trialing approaches for increasing diversity and inclusion at events, including
workshops and conferences. Gathering statistics on the relative success of
these approaches and their contribution to improving diversity within the
computational science developer community.
Looking at opportunities to further increase “domain mobility” through tak-
ing advantage of existing courses/training material or offering new informa-
tion that can help individuals move more practically between working in
different domains. Look at take-up of such opportunities across different do-
mains and use this as a basis for longer-term analysis on how diversity in
the RSE community changes over time.
Once individuals become RSEs and engage with the RSE community, the
RSE community meets many of the requirements to ensure that it welcomes
and supports diversity. The open, inclusive, nature of the community is of great
importance here and, using gender diversity as an example, there are already
several women in highly-visible leadership roles.
Developing software for science and the wider computational research domain
is challenging. The relatively new field of research software engineering encom-
passes many of the individuals undertaking these software development tasks,
regardless of whether their official job title considers them to be a researcher,
professor, software engineer or RSE. It is clear that research software engineer-
ing, and research software in general, shares many diversity challenges with the
wider software development field but we feel there are many opportunities to ad-
dress this. This begins with increasing awareness of the different aspects raised
in this paper and using this as a basis to develop the environment and oppor-
tunities to help build equity, diversity and inclusion within the community. We
hope that the research software community continues to grow, and become more
diverse by sharing and learning from other practitioners.
5 Acknowledgement
We thank Simon Hettrick, James Graham and Rob Haines for their input, and
Olivier Philippe who originally processed the international RSE survey data.
NCH and CJ acknowledge support from EPSRC / BBSRC / ESRC / NERC /
AHRC / STFC / MRC grant EP/S021779/1 for the UK Software Sustainability
Institute. JC acknowledges support from EPSRC grant EP/R025460/1.
Understanding EDI Challenges Within the Research Software Community 13
References
1. Apa, A.: Driving Gender Equality in the Tech Industry: Breaking down un-
conscious bias (Jun 2018), [Online] Available at: https://womeninhpc.org/hpc/
driving-gender-equality
2. Bano, M., Zowghi, D.: Gender disparity in the governance of software engineering
conferences. 2019 IEEE/ACM 2nd International Workshop on Gender Equality in
Software Engineering (GE) (May 2019). https://doi.org/10.1109/ge.2019.00016
3. Brett, A., Croucher, M., Haines, R., Hettrick, S., Hetherington, J., Stillwell, M.,
Wyatt, C.: Research Software Engineers: State of the Nation Report 2017 (April
2017). https://doi.org/10.5281/zenodo.495360
4. Campbell, L.G., Mehtani, S., Dozier, M.E., Rinehart, J.: Gender-heterogeneous
working groups produce higher quality science. PLoS ONE 8(10), e79147 (Oct
2013). https://doi.org/10.1371/journal.pone.0079147
5. Capretz, L.F., Ahmed, F.: Why do we need personality diversity in software en-
gineering? ACM SIGSOFT Software Engineering Notes 35(2), 1–11 (Mar 2010).
https://doi.org/10.1145/1734103.1734111
6. Chamorro-Premuzic, T.: How to design a diversity intervention that actually works
(Jun 2019), [Online] Available at: https://www.forbes.com/sites/tomaspremuzic/
2019/06/20/how-to-design- a-diversity-intervention-that-actually-works/
7. Cheryan, S., Plaut, V.C., Davies, P.G., Steele, C.M.: Ambient belong-
ing: How stereotypical cues impact gender participation in computer sci-
ence. Journal of Personality and Social Psychology 97(6), 1045–1060 (2009).
https://doi.org/10.1037/a0016239
8. Chue Hong, N., Cohen, J., Jay, C.: RSE Diversity Analysis Notebooks (Apr 2021).
https://doi.org/10.5281/zenodo.4662166
9. Collis, T.: Improving diversity at hpc conferences and events (2016), J. West, L.
Rivera and N. Chue Hong (eds.). [Online] Available at http://www.hpc-diversity.
ac.uk/best-practice-guide/access-best-practice
10. Favaro, B., Oester, S., Cigliano, J.A., Cornick, L.A., Hind, E.J., Parsons, E.C.M.,
Woodbury, T.J.: Your science conference should have a code of conduct. Frontiers
in Marine Science 3(Jun 2016). https://doi.org/10.3389/fmars.2016.00103
11. Foxx, A.J., Barak, R.S., Lichtenberger, T.M., Richardson, L.K., Rodgers, A.J.,
Webb Williams, E.: Evaluating the prevalence and quality of conference codes of
conduct. Proceedings of the National Academy of Sciences 116(30), 14931–14936
(Jul 2019). https://doi.org/10.1073/pnas.1819409116
12. Freeman, R.B., Huang, W.: Collaborating with People Like Me: Ethnic Coauthor-
ship within the United States. Journal of Labor Economics 33(S1), S289–S318 (Jul
2015). https://doi.org/10.1086/678973
13. Gardiner, M., Aurora, V., Smith, S., Benjamin, D.: Geek Feminism Wiki: Con-
ference anti-harassment/Policy, [Online] Available at: https://geekfeminism.wikia.
org/wiki/Conference anti-harassment/Policy, Accessed 10th Aug 2020.
14. Graham, H., Raeside, R., Maclean, G.: Understanding the status of under-
represented groups in the information and communication technologies: A report
to the engineering and physical sciences research council (July 2017), [Online]
Available at https://epsrc.ukri.org/newsevents/pubs/napierdiversityreport/
15. HESA: HE staff by activity standard occupational classification, atypical marker,
contract marker, mode of employment, terms of employment, contract levels, sex,
age group, disability, ethnicity and academic year, [Online] Available at: https:
//www.hesa.ac.uk/data-and-analysis/staff/table-3
14 N.P. Chue Hong et al.
16. HESA: HE student enrolments by subject area and sex, [Online] Available at:
https://www.hesa.ac.uk/data-and-analysis/sfr247/figure- 14
17. Institute of Physics: What does a physicist look like? (Mar 2016), [Online] Available
at: http://www.iop.org/publications/iop/2016/file 67244.pdf.
18. Isbell, L.A., Young, T.P., Harcourt, A.H.: Stag parties linger: Continued gender
bias in a female-rich scientific discipline. PLoS ONE 7(11), e49682 (Nov 2012).
https://doi.org/10.1371/journal.pone.0049682
19. Liang, T., Liu, C., Lin, T., Lin, B.: Effect of team diversity on software project per-
formance. Industrial Management & Data Systems 107(5), 636–653 (May 2007).
https://doi.org/10.1108/02635570710750408
20. Mathieson, S.A.: How diversity spurs creativity in software development
(Dec 2019), [Online] Available at: https://www.computerweekly.com/feature/
How-diversity-spurs-creativity-in-software-development. Accessed 26th Jan 2021.
21. Merritt, C.: Why do we keep hiring ourselves? (Sep 2018), [On-
line] Available at: https://womeninhpc.org/diversity-and-inclusion/
why-do-we-keep-hiring-ourselves
22. NCWIT: What is the impact of gender diversity on technology business perfor-
mance? (2014), [Online] Available at: https://www.ncwit.org/sites/default/files/
resources/impactgenderdiversitytechbusinessperformance print.pdf.
23. NumFOCUS: NumFOCUS DISCOVER Cookbook (Diverse & Inclusive Spaces and
Conferences: Overall Vision and Essential Resources) (2019), [Online] Available at:
https://discover-cookbook.numfocus.org/. Accessed 13th Aug 2020.
24. O. Philippe et al.: softwaresaved/international-survey: Public release for 2018 re-
sults (Mar 2019). https://doi.org/10.5281/ZENODO.2585783
25. Society, B.C.: BCS Diversity Report 2020: ONS Analysis (Jun 2020), [Online]
Available at: https://www.bcs.org/media/5766/diversity-report- 2020-part2.pdf.
26. Stahl, G.K., Maznevski, M.L., Voigt, A., Jonsen, K.: Unraveling the effects of
cultural diversity in teams: A meta-analysis of research on multicultural work
groups. Journal of International Business Studies 41(4), 690–709 (Nov 2009).
https://doi.org/10.1057/jibs.2009.85
27. Sufi, S., Jay, C.: Raising the status of software in research: A survey-based
evaluation of the software sustainability institute fellowship programme [version
1; peer review: 3 approved with reservations]. F1000Research 7(1599) (2018).
https://doi.org/10.12688/f1000research.16231.1
28. The Hopper Fund: The Hopper Conference Diversity Guide: Do Better at Confer-
ence Diversity, [Online] Available at: https://conference.hopper.org.nz/. Accessed
13th Aug 2020.
29. Tulloch, A.I.T.: Improving sex and gender identity equity and inclusion at con-
servation and ecology conferences. Nature Ecology & Evolution 4(10), 1311–1320
(Aug 2020). https://doi.org/10.1038/s41559-020-1255-x
30. Universities UK: Patterns and trends in UK higher education 2018 (Sep
2018), [Online] Available at: https://www.universitiesuk.ac.uk/facts-and-stats/
data-and-analysis/Documents/patterns-and- trends-in-uk- higher-education-2018.
pdf.
31. Van Der Zee, K., Atsma, N., Brodbeck, F.: The influence of social identity and per-
sonality on outcomes of cultural diversity in teams. Journal of Cross-Cultural Psy-
chology 35(3), 283–303 (May 2004). https://doi.org/10.1177/0022022104264123
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Conferences are important for professional learning and for building academics’ reputations and networks. All members of the scientific community should feel supported and free to contribute their best at such events. I evaluated the actions and policies of conferences held by international academic societies for ecology and conservation since 2009, to assess whether conferences support participation across sexual orientations and gender identities. Although half of the 30 conferences had codes of conduct promoting equity, diversity and inclusion, the quantity and quality of initiatives to support such principles varied between societies and years. Conferences with codes were significantly more likely to implement structural initiatives to minimize discrimination or harassment, such as procedures for reporting misconduct and submission guidelines to promote speaker diversity, as well as initiatives to support parents. Initiatives minimizing barriers to attendance were rare; 47% of conferences were held in locations that discriminate against certain identities and <10% promoted event safety and accessibility to potential attendees. Piecemeal actions and inadequate conference evaluation have resulted in no significant growth in gender equity or diversity initiatives. I propose a six-step timeline that improves conference inclusion by embedding diversity and equity into planning, financing, marketing, scientific and social scheduling, evaluation and reporting.
Article
Full-text available
Background: This paper reports the results of an evaluation of the Software Sustainability Institute’s Fellowship Programme, which focused on identifying and categorising the benefits that the fellowship has afforded its recipients, via a series of open questions. Methods: The evaluation took the form of a survey open to people awarded Fellowships between 2012 and 2016, which asked people to report the effect that the programme had had on them, their institutions, their research domains and their careers. Results: The results show that the Fellowship plays a wide-ranging role in supporting communities of best practice and skills transfer, and that a significant benefit is the way it has raised the profile of software in research, and those people who develop and advocate for it. Conclusions: The evaluation of the programme has shown the need to support research software in situ and credit the engineers and researchers who are working in this important area that supports reproducibility, reuse and the integrity of research investments.
Article
Full-text available
By examining the ethnic identity of authors in over 2.5 million scientific papers written by US-based authors from 1985 to 2008, we find that persons of similar ethnicity coauthor together more frequently than predicted by their proportion among authors. The greater homophily is associated with publication in lower-impact journals and with fewer citations. Meanwhile, papers with authors in more locations and with longer reference lists get published in higher-impact journals and receive more citations. These findings suggest that diversity in inputs by author ethnicity, location, and references leads to greater contributions to science as measured by impact factors and citations.
Article
Full-text available
Here we present the first empirical evidence to support the hypothesis that a gender-heterogeneous problem-solving team generally produced journal articles perceived to be higher quality by peers than a team comprised of highly-performing individuals of the same gender. Although women were historically underrepresented as principal investigators of working groups, their frequency as PIs at the National Center for Ecological Analysis and Synthesis is now comparable to the national frequencies in biology and they are now equally qualified, in terms of their impact on the accumulation of ecological knowledge (as measured by the h-index). While women continue to be underrepresented as working group participants, peer-reviewed publications with gender-heterogeneous authorship teams received 34% more citations than publications produced by gender-uniform authorship teams. This suggests that peers citing these publications perceive publications that also happen to have gender-heterogeneous authorship teams as higher quality than publications with gender uniform authorship teams. Promoting diversity not only promotes representation and fairness but may lead to higher quality science.
Article
Full-text available
Discussions about the underrepresentation of women in science are challenged by uncertainty over the relative effects of the lack of assertiveness by women and the lack of recognition of them by male colleagues because the two are often indistinguishable. They can be distinguished at professional meetings, however, by comparing symposia, which are largely by invitation, and posters and other talks, which are largely participant-initiated. Analysis of 21 annual meetings of the American Association of Physical Anthropologists reveals that within the subfield of primatology, women give more posters than talks, whereas men give more talks than posters. But most strikingly, among symposia the proportion of female participants differs dramatically by the gender of the organizer. Male-organized symposia have half the number of female first authors (29%) that symposia organized by women (64%) or by both men and women (58%) have, and half that of female participation in talks and posters (65%). We found a similar gender bias from men in symposia from the past 12 annual meetings of the American Society of Primatologists. The bias is surprising given that women are the numerical majority in primatology and have achieved substantial peer recognition in this discipline.
Article
Full-text available
Purpose – This research seeks to investigate the relationship between knowledge diversity (KD) in software teams and project performance. Previous research has shown that member diversity affects team performance; most of that work, however, has focused on diversity in personal or social attributes, such as gender or social category. Current research targets at the knowledge level aim to facilitate the implementation of knowledge management in organizations. Design/methodology/approach – A research framework was developed based on conflict theory and empirically tested on software teams in Taiwan. Findings – It was found that KD increases task conflict, which in turn has significant positive effects on team performance and that value diversity (VD) increases relationship conflict, which in turn negatively affects team performance. Research limitations/implications – The findings indicate that task conflict can enhance team performance, while relationship conflict can reduce team performance. Therefore, it is important to maintain healthy relationships among team members. Practical implications – This research concludes that KD is beneficial and that VD is harmful to project outcome in software development. It is, therefore, useful for managers to form teams whose members encompass a broad knowledge base. Originality/value – This paper proposes a novel way to measure knowledge and VD in teams and reports the effects of these attributes on team performance. The work also shows that a proper level of task conflict in a software team is necessary for achieving high performance.
Article
Efforts to increase inclusion in science face multiple barriers, including cultural and social behaviors in settings such as academic conferences. Conferences are beneficial, but the culture can promote inequities and power differentials that harm historically underrepresented groups. Science suffers when conference culture propagates exclusion and discrimination that leads to attrition of scientists. Codes of conduct represent a tool to shift conference culture to better support diverse scientists and clearly detail unacceptable behaviors. We examined the prevalence and content of codes of conduct at biology conferences in the United States and Canada. We highlight how codes of conduct address issues of sexual misconduct and identity-based discrimination. Surprisingly, only 24% of the 195 surveyed conferences had codes. Of the conferences with codes, 43% did not mention sexual misconduct and 17% did not mention identity-based discrimination. Further, 26% of these conferences failed to include a way to report violations of the code and 35% lacked consequences for misconduct. We found that larger and national conferences are more likely to have codes than smaller ( P = 0.04) and international or regional ( P = 0.03) conferences. Conferences that lack codes risk creating and perpetuating negative environments that make underrepresented groups feel unwelcome, or worse, actively cause harm. We recommend that conferences have codes that are easily accessible, explicitly address identity-based discrimination and sexual misconduct, provide channels for anonymous impartial reporting, and contain clear consequences. These efforts will improve inclusivity and reduce the loss of scientists who have been historically marginalized.
Conference Paper
In this paper, we discuss gender disparity in software engineering (SE) conferences. We have examined the roles of General Chair, Program Chair, and main track Program Committee members in six highly ranked conferences in SE for a period of ten years in order to understand the pattern of gender disparity in visible roles. We also present the opinions elicited from ten participants on this topic, who have served at some of these SE conferences in leadership roles. Our aim is to reflect on the current state and initiate the debate, on gender equality in SE conferences.
Article
The present study examined the influence of social identity and personality on work outcomes among business students who worked together in culturally diverse teams. As predicted, a negative effect of identification with one’s cultural background and a positive effect of identification with the team on well-being were found under conditions of high diversity. For commitment, the same pattern of findings was obtained, but now the impact of identification with the team was found regardless of the level of diversity. No support was found for strong positive outcomes associated with the case in which individuals identify with the team and with their cultural background.With respect to personality, the intercultural traits of Emotional Stability and Flexibility were found to have a positive effect on work outcomes under conditions of high diversity. Interestingly, whereas Flexibility had a positive effect on exam grades under conditions of high diversity, a negative effect of this trait was found under conditions of low diversity.