In this article, using Github data, claimed to be the largest software repository, gender bias in the open-source software community is investigated by analyzing the acceptance rate of pull requests submitted by female and male developers, which the gender is obtained through a novel gender linking technique using contributors’ social profiles. Although the results demonstrate the women’s high successful contribution, pieces of evidence have been put forward that bias against women still exists in the evaluated community. The research mainly intends to answer an analytical question: ‘To what extent does gender bias exist when pull requests are judged on Github? ’
To this end, researchers attempt to address if women’s pull requests are less likely to be accepted than men’s, in terms of several effects. Examining the status of all pull requests, results show an opposite direction of the initial assumption; women's pull requests are significantly more accepted than those of men. Since the results seem not to be in line with the hypothesis of gender bias in the field, researchers explain the results by decomposing it into evaluating different factors; the acceptance of women’s pull requests and its trend over time; risk-taking behavior, the focal point of attention, importance, and size of women’s pull requests; the complexity/readability of changes made in a variety of programming languages; and eventually, a sociology point of view as favoring women’s pull requests intending to help.
Forasmuch as the results illustrate women’s pull request acceptance rate outpace men’s in almost all factors, observations draw researchers’ attention to a controversial decision to clarify even the possibility of reverse discrimination, and that is re-analyzing the factors by controlling covariates, keeping the similar data in both gender groups and removing the outliers, which to some extent acts as a data scrubber. While the statistical power is reduced because of the matching procedure, it controls the notable differences observed in light of mentioned factors.
According to the results gained by reconsidering of factors in the reduced sample, it is shown that women’s pull requests acceptance rate drops when their gender is apparent despite earlier proof of their high-quality performance, which can be considered as evidence for the existence of bias against women in open source. Turning to inevitable results obtained in this study, women seem to be more competent than men in male-dominated occupations, which can be explained through different theories.
In this article, using
Github
data, claimed to be the largest software repository, gender
bias
in the open-source software community
is investigated
by analyzing the
acceptance
rate of
pull
requests
submitted by female and male developers, which the gender
is obtained
through a novel gender linking technique using contributors’ social profiles. Although the
results
demonstrate the
women’s
high successful contribution, pieces of evidence have
been put
forward that
bias
against
women
still
exists in the evaluated community. The research
mainly
intends to answer an analytical question: ‘To what extent does gender
bias
exist when
pull
requests
are judged
on
Github
? ’
To this
end
, researchers attempt to address if
women’s
pull
requests
are less likely to be
accepted
than
men’s
, in terms of several effects. Examining the status of all
pull
requests
,
results
show
an opposite direction of the initial assumption; women's
pull
requests
are
significantly
more
accepted
than those of
men
. Since the
results
seem not to be in line with the hypothesis of gender
bias
in the field, researchers
explain
the
results
by decomposing it into evaluating
different
factors
; the
acceptance
of
women’s
pull
requests
and its trend over time;
risk
-taking behavior, the focal point of attention, importance, and size of
women’s
pull
requests
; the complexity/readability of
changes
made in a variety of programming languages; and
eventually
, a sociology point of view as favoring
women’s
pull
requests
intending to
help
.
Forasmuch
as the
results
illustrate
women’s
pull
request
acceptance
rate outpace
men’s
in almost all
factors
, observations draw researchers’ attention to a controversial decision to clarify even the possibility of reverse discrimination, and
that is
re-analyzing the
factors
by controlling
covariates
, keeping the similar data in both gender groups and removing the outliers, which to
some
extent acts as a data scrubber. While the statistical power is
reduced
because
of the matching procedure, it controls the notable differences observed in light of mentioned factors.
According to the
results
gained by reconsidering of
factors
in the
reduced
sample, it
is shown
that
women’s
pull
requests
acceptance
rate drops when their gender is apparent despite earlier proof of their high-quality performance, which can
be considered
as evidence for the existence of
bias
against
women
in open source. Turning to inevitable
results
obtained in this study,
women
seem to be more competent than
men
in male-dominated occupations, which can be
explained
through
different
theories.