Mentorship of Women in OSS Projects: A Cross-Disciplinary, Integrative Review

Abstract

Mariska Jacobs, Reed Milewicz, and Alexander Serebrenik

Mentorship is a relationship in which a more experienced or more knowledgeable person (a mentor) helps guide a less experienced or less knowledgeable person (a mentee); here we draw upon mentor role theory, which holds that mentorship often acts as a vehicle for both career development and psychosocial support, with the end goal of cultivating the mentee’s whole self [14]. While mentorship can benefit everyone, studies have shown that positive mentorship experiences are especially significant for members of underrepresented groups. A 2019 report by the National Academies of Sciences, Engineering, and Medicine (NASEM) found that mentorship can be a powerful tool for cultivating professionals from underrepresented groups in STEMM fields [7]; through a close working alliance with a mentor, women and minority mentees can acquire not just the skills they need to succeed but also an affirmation of belonging and professional identity that is so crucial to retention.¹

In this chapter, we examine the practice of mentorship as a strategy for the retention of women in OSS contexts. Scholarship in recent years has drawn attention to the importance of diversity, equity, and inclusion in growing and sustaining open source software (OSS) communities [SE6]. Along gender lines, women developers are underrepresented in OSS projects and have a higher disengagement rate than men [17]. Studies suggest that a mix of factors such as gender bias [SE15] and a lack of confidence [SE30] affect rates of early disengagement; as Singh and Bongiovanni put it, women in OSS must engage in “vexatious labor” to prove their worth and maintain their place in the community [18]. One interpretation of the data is that women contributors to OSS seek a “sense of belonging” that is held at risk by barriers to inclusion [13, 19, 22].

While mentorship has been shown to be an effective mechanism for onboarding support in OSS development [3, 9, 11, 21], the connection between mentorship practices and diversity and inclusion (D&I) has been understudied in software engineering. Numerous studies outside of software engineering, however, have linked mentorship with diversity and inclusion [5, 6, 7, 8, 10, 24]. Here we see an opportunity to leverage insights from the broader literature on mentorship to help frame the challenges faced by women contributors to OSS and how inclusive mentorship practices may help address those challenges.

To that end, we present findings from a cross-disciplinary review of mentorship among women professionals in OSS contexts. Following the guidelines by Petersen et al. [16], we investigated the challenges women protégées face in OSS and possible strategies that may help overcome those challenges. To further enrich our understanding of these barriers and strategies, we conducted a second mapping study on the challenges faced by women in mentorship in the broader literature outside software engineering. This approach enables us (1) to situate OSS mentorship in the broader context, (2) to compare and contrast the experiences of women professionals across disciplines, and (3) to surface additional strategies for ensuring the retention of women in OSS through mentorship.

Background

There is a growing body of evidence pointing toward mentorship as a key mechanism for onboarding and retention in open source communities, and researchers in recent years have sought to understand the practice of mentorship in OSS – especially as it concerns newcomers. Fagerholm et al. [9] examined how mentoring and project characteristics influence the effectiveness and efficiency of onboarding. Balali et al. [2, 3] and Steinmacher et al. [SE26] have investigated mentorship in OSS, the challenges the mentors and mentees face, and the strategies they use to navigate those challenges. While formal mentorship programs are less common outside of large open source communities, findings from Feng et al. suggest that informal mentorship relationships between junior and senior contributors are widespread in OSS projects [SE11].

With regard to women in OSS, Balali et al. have called attention to gender-specific challenges faced by women receiving mentorship, including confidence issues, differences in communication styles, and underestimation of their skills by their peers [3]. Work by Ford et al. arrived at similar findings in a study of the Stack Overflow community [12]; the authors found that women are more likely than men to doubt that they have the level of expertise needed to contribute and to feel overwhelmed when competing with a large number of other contributors and recommended mentorship as a strategy for helping address them. Meanwhile, a recent literature survey by Trinkenreich et al. on women’s participation in OSS recommends training mentors to better guide women newcomers [23]. Singh et al. highlight a need for more same-gender mentoring between women in OSS and that communities should reward and recognize women who help onboard and mentor others [18].

We note that the potential for mentorship to create a more inclusive climate for women professionals is not unique to OSS. Indeed, while mentorship in software engineering has received increasing attention in recent years, this scholarship represents only a small fraction of the broader mentorship literature (cf., a cross-discipline analysis of mentorship by Lefebvre et al. [15]), including literature on mentorship and diversity and inclusion (D&I); we see opportunities to leverage that wealth of knowledge to inform mentorship practice in our domain. As noted by Wagner et al., for topics that span different disciplines, “the discovery of different methodologies, operationalizations, constructs, or relationships from other disciplines may inspire further research and ultimately broaden and enrich the understanding” of a given topic [25]. For mentorship in particular, Allen and Eby recommend taking a cross-disciplinary view as this allows “researchers and practitioners to obtain a richer and more inclusive perspective of the primary themes of mentoring research and practice” [1].

Methodology

The goal of our work is to identify challenges mentors and women mentees face in OSS contexts and possible strategies that may help overcome those challenges. To do this, we performed a cross-disciplinary twin systematic mapping study. That is, using established guidelines [16], we conducted two mapping studies (see Figure 20-1): the first targeting OSS mentorship literature and the second targeting comparable literature on mentorship of women in other disciplines – the latter is an offshoot of a broader, ongoing systematic literature review on the meaning of mentorship across disciplines. Using these results, we synthesized a systematic map (Figure 20-2) to provide an overview of the best available evidence on this subject. The following research questions guided the study:

RQ1 What are the challenges that women mentees face in an OSS project? How do these challenges compare to those faced by women mentees in other disciplines?

RQ2 What strategies can help overcome the challenges faced by mentors and women mentees in OSS projects? Are there additional strategies from the broader mentorship literature that could be applicable to OSS contexts?

We opted to use Google Scholar because it provides a single point of access to multiple data sources both inside and outside of software engineering and its support for Boolean expressions. After several rounds of iterative refinement, we selected the following search strings

Search	Query String
(S1) Mentorship of Women in OSS Contexts	mentoring AND (onboarding OR challenges OR barriers) AND (female OR women OR gender) AND (OSS OR "open source software")
(S2) Mentorship in Other Disciplines	mentorship AND (mentorship OR mentoring OR mentor*) AND (characteristics OR qualities OR outcomes OR barriers OR facilitators)

together with the following inclusion (IC) and exclusion criteria (EC):

Figure 20-1

An illustration of the methodology used to collect and assess evidence from scholarly literature on the mentorship of women

IC1 The study presents an approach to mentoring women either in an OSS project or similar organizational contexts.

IC2 The study describes challenges or barriers of women in an OSS project or similar organizational contexts.

IC3 The study evaluates mentoring women or newcomers in an OSS project or similar organizational contexts.

IC4 The abstract of a study mentions both women and mentorship, and from the abstract it is clear that the paper can be used to answer one of our research questions.

EC1 The study is not peer-reviewed.

EC2 The study is not written in English.

We executed our search on OSS mentorship on December 14, 2020, and the search on mentorship more broadly on April 10, 2022, each returning the top 1,000 search results mined from Google Scholar. The papers were evaluated following the Google Scholar ranking against the inclusion and exclusion criteria. For each corpus, the authors used open coding techniques to extract concepts and topics surrounding barriers to and strategies for the mentorship of women and constructed a mapping to organize what is known. Of note, when analyzing studies on mentorship outside of software engineering, we coded both claims directly made by studies and claims alluded to in sources cited; for example, [GL22] examines the mentorship of women in policing and cites other relevant sources on the experiences of women in that profession. This allows us to capture the state of practice across different disciplines and their respective bodies of knowledge.

Results

In total, 43 papers met the inclusion criteria for mentorship of women in OSS (S1), and 53 papers met the inclusion criteria for mentorship of women in other professional, non-SE contexts (S2). The papers on OSS mentorship discuss both formal [SE19] and informal mentoring [SE16], mentoring in Apache projects [SE3] and via summer programs such as Google Summer of Code [SE25], mentoring programs designed for women and minority groups such as Outreachy and Rails Girls Summer of Code [SE25], as well as mentoring programs targeting all newcomers. Most of the papers have been published between 2015 and 2020, in venues targeting software engineering (e.g., ICSE, FSE) and collaborative work (e.g., CSCW). The papers in our cross-disciplinary corpus, meanwhile, cover the mentorship of women in diverse professional contexts, including higher education [GL6], primary and secondary school teaching [GL18], executive boardrooms [GL4], health and fitness [GL15], medicine [GL24], and policing [GL22]. These papers were published between 2007 and 2021 and were spread evenly over a wide range of venues.

The results of the mapping study on these corpora are summarized in Figure 20-2. With respect to the challenges that women mentees face in seeking mentorship, we grouped these into personal, relational, and organizational barriers. With respect to strategies that can help, there are two broad categories: (1) adopting mentorship practices that make mentorship more accessible and effective (e.g., encouraging mentees having multiple mentors) or that may particularly benefit women mentees (e.g., same-gender mentorship) and (2) policies that can be implemented at a project or organizational level to build effective routines for inclusive mentoring (e.g., training and monitoring mentorship progress). A third category (Other Strategies) was added to capture innovative ideas that did not neatly fit into the first two groups. Here, we weave together insights from the OSS literature with comparable evidence on the mentorship of women in other disciplinary contexts in order to provide a richer perspective on the challenges women face and concrete methods for overcoming those challenges.

Figure 20-2

Our systematic map of challenges faced by women mentees and strategies to overcome those challenges based on both our cross-disciplinary and OSS-specific corpora. Barriers and strategies that were attested in the OSS literature are labeled.

RQ1. What Are the Challenges That Women Mentees Face in Receiving Mentorship in Professional/OSS Contexts?

(Personal) Women are seen and/or see themselves as less capable: A recurring theme among the studies in our cross-disciplinary corpus is that women are frequently both seen and may see themselves as less capable than men. In male-dominated industries, women are often viewed as outsiders whose competencies are – by default – in question [GL13, GL21]. This is especially true for women who belong to multiple minority groups, such as women of color in higher education, who frequently find themselves “having their authority challenged while having their qualifications and intelligence questioned” [GL5]. As a result, women professionals may be reluctant to seek mentorship for fear of appearing incompetent or in need of help [GL7].

In OSS: Bias against women and negative assumptions about their ability to contribute are a prominent factor driving the underrepresentation of women in OSS [SE15]. Terrell et al. [SE28] showed that the acceptance rate of pull requests of women is higher than that of men when women hide their gender, but lower when they reveal their gender. Women contributors may internalize these negative perceptions; Balali et al. found that women’s sense of self-efficacy is lower than men’s and that women feel less comfortable and less accepted by male counterparts [SE3]. It is worth noting, however, that Bosu and Sultana found that gender bias is not always prominent in OSS projects [SE6]. Studying gender diversity and inclusion in OSS projects through analysis of code review repositories, their results show that three out of the ten projects indicate technical biases against women developers, including lower code acceptance rates as well as delayed feedback during code reviews for women.

(Personal) Women are perceived as competing priorities in the home: Women in the workforce frequently face countervailing demands in the home and at work that can affect how they perceive mentorship. Ghosh and Haynes observe that in many cultures “women are taught to think of their career aspirations to be secondary and to give family priority over work” [GL13]. As an example of this, Leck et al. note that “while male protégés may perceive outcomes of mentoring as valuable (e.g., getting a promotion), women may not because promotions are typically accompanied by increased responsibility which may impinge on other activities women value, such as child care and family time” [GL7]. Mismatches in priorities and expectations around mentorship on both sides can lead to less successful and productive mentor-mentee relationships.

In OSS: We did not find evidence for gendered expectations around work life and home life in OSS contexts as in our cross-disciplinary review. We hypothesize that this may be due to several factors, including the degree and duration of involvement in OSS projects (i.e., women’s careers in the workforce play out over their entire life course vs. involvements in OSS projects, which may be more episodic) and/or demographic differences (i.e., with respect to age and family status). There is some evidence, however, to suggest that women may express different motivations to join OSS projects compared with men. Balali et al. found that women contributors were more likely to report joining an OSS project out of excitement rather than a sense of obligation [SE3]. Barcomb studied retaining voluntary episodic contributors in OSS software communities and thereby took gender as one of the moderating effects [SE5]. Although not supported with significance due to a small number of women in the sample size, results suggest large differences between men and women; women had higher path coefficients for contributor benefit motives, social norms, psychological sense of community, and community commitment and a lower path coefficient for satisfaction than men.

(Personal) Women have to adapt to male standards of behavior: As a result of stereotypical gender roles – both those imposed and internalized – women professionals who want to get ahead of their careers often must change how they behave. Jones cites numerous sources that show how women frequently “mask their femininity or behave in masculine ways to deflect unwelcome attention, blend into the system and to achieve success” [GL22]. Case in point, women in many cultures are expected to be more passive and compliant than men, which may make women hesitant to speak up and ask for mentorship [GL2]. Likewise, Fowler notes that women professionals are frequently expected to say “yes” to work responsibilities because of the negative ramifications of saying “no” [GL3]; the author recommends that women mentees may benefit from career advice on how to navigate those conversations. Due to differences in communication styles between men and women, studies suggest that women are more likely to adjust their communication patterns to avoid misinterpretation and miscommunication, including in mentor-mentee contexts [GL1].

In OSS: Studies in OSS contexts have also identified differences in communication styles between men and women and expectations of conduct placed upon women as potential barriers. Imtiaz et al. comment on the “tightrope effect,” where women can only express a limited range of behaviors deemed socially acceptable [SE15]; being too polite or impolite has bigger negative consequences for women than for men, as men are often forgiven for their behavior and sometimes even rewarded for being brash. Gallus and Bhatia found significant gender differences in people’s conversational styles when studying the role of gender in Wikipedia contributions [SE14]; the authors show, however, that differences in communication styles diminish when women reach positions of authority, illustrating how women must change their behaviors as they “climb the ladder.” Meanwhile, Balali et al.’s study [SE3] shares reports of women on the differences in communication style and these causing men to come off as creepy.

(Relational) Inappropriate behavior from mentors and/or perceptions of impropriety: Because mentorship traditionally involves a more senior person guiding a less experienced junior, the power dynamics can make mentees vulnerable to mistreatment or exploitation. For example, mentees may experience bullying or incivility from their mentors and not be in a position to push back [GL11]. These issues can be especially pronounced for women mentees. In fields where men outnumber women, mentor-mentee relationships are frequently cross-gender, which can raise suspicions of sexual or romantic impropriety [GL7]; concerns about how the relationship may be perceived may lead to mentors or mentees limiting their one-on-one interactions [GL13]. Of importance to our work, Leck et al. cite findings from multiple sources that online mentoring (or e-mentoring) may reduce the impact of differences of age, gender, and race in mentor-mentee relationships [GL16].

In OSS: In OSS contexts, studies have observed how harassment and sexually charged environments can drive away women participants. Nafus highlights how sexualized online environments include “willful inattention to offensive talk,” with one interviewee describing how male developers casually joked about women being raped [SE20]. Lin and den Besten [SE17] demonstrate that, in the open source development, languages or presuppositions or anecdotes cited for strategic communication can reflect certain types of cultures and they dominate the stage/field/platform. Humor is often used to create common ground and gain the attention and interest of people who might have the key to solving a problem. Unfortunately, what looks appealing to some groups will offend others. In a male-dominated environment, male humor dominates the field and tends to exclude consideration of women’s needs, values, and preferences. Meanwhile, a 2010 study by Powell et al. found that 50% of women contributors to OSS reported having witnessed gender-based harassment within the OSS community and that 50% experienced harassment themselves – 62% saying that harassment was a deterrent to participation in OSS [SE22]. To the best of our knowledge, the extent to which such project cultures affect the mentorship of women in OSS contexts has not been specifically studied, but the findings in our cross-disciplinary corpus would suggest that women would be less likely to seek mentorship from men in such environments. In the interviews conducted by Balali et al., participants noted how women mentees would have conversations with women mentors that they probably would not have had with men mentors [SE3]; this can be a problem because the success of mentoring decreases when mentees feel like they cannot share all their concerns with their mentors.

(Relational) Mentors reluctant to mentor women: Several papers in our cross-disciplinary corpus reported mentors being reluctant to support women mentees due to in-group/out-group biases and/or concerns about underperforming mentees harming the social status of their mentors. McDonald and Westphal found that men in corporate executive roles provided relatively more mentorship to men who were first-time directors compared with women and suggested that this may be due to in-group/out-group bias as women may be seen as outsiders or less relatable [GL4] (see also [GL12]). Women mentors, meanwhile, may be judged by how well their mentees perform, leading them to be more selective in whom they mentor; in a study of sports mentorship, Harden et al. observe “female mentors in leadership position are concerned about their reputation if a protégé makes a mistake. Essentially, the female mentor believes this has a negative impact on her ability as a leader” [GL14].

In OSS: In the context of OSS, recent work by Feng et al. found that 93.81% of implicit mentoring interactions in OSS projects were between same-gender pairs and that women were more significantly more likely to participate in cross-gender mentoring than men [SE11]. The authors suggest that unconscious bias and the tendency for people to preferentially interact with others like themselves may explain why men mentors tend to mentor other men.

(Relational) Women less likely to be included and less likely to receive mentorship: In our cross-disciplinary corpus, we identified a recurring theme of women being less likely to be included in workplace activities, limiting their opportunities to find and establish relationships with potential mentors. A review of mentorship in higher education by Fowler finds that “women reported feeling less included in discussions about research, teaching, and promotion and three times less likely to receive career help from colleagues than men […] Such exclusion, whether intentional or not, is a major barrier to achievement” [GL3] (see also [GL11]). In policing, Jones argues that findings such as these suggest that “compared to their male counterparts, women may have fewer formal and informal opportunities for developing mentoring relationships” [GL22]. According to Kalpazidou and Faber, having fewer opportunities to network and learn from others may be a driving factor in women professionals exiting their careers [GL10].

In OSS: For OSS projects, it is crucial to adequately integrate newcomers into projects, but resources for mentorship can be limited. Balali et al. have observed that a lack of time for mentoring is a significant barrier for mentees to receive mentorship in general [SE3]. Meanwhile, Fiesler et al. note there is a perception that mentorship responsibilities can distract mentors from doing core work on projects [SE12]. Women contributors may be particularly sensitive to insufficient support during onboarding.

(Organizational) Male-gendered environments disadvantage women in mentorship: For the purposes of our work, we follow the definition of male-gendered industries given by Ramaswami et al. [GL21]: industries where either (1) men make up 75% or more of the workforce or (2) the profession is male-stereotyped (e.g., aggressive, competitive, engineering-intensive, etc.). Studies suggest that the culture, norms, and practices of male-gendered professions tend to put “outsiders” (women and minorities) at a disadvantage. In medicine, McNamara et al. argue, “Mentoring challenges faced by women in health care may arise from a clash between social roles and a male-dominated hierarchical medical culture” [GL2]. Similar findings are reported by Kalpazidou and Faber in higher education, where women researchers frequently report a lack of culture fit in academia and a greater sense of isolation relative to men [GL10].

In OSS: The OSS community fits with the definition of a male-gendered working environment. Not only do women make up a small minority of OSS contributors, the disparities are especially pronounced among top contributors; Wang et al. found that only 3% of the top 5,000 developers on GitHub are women [SE30]. Using our cross-disciplinary corpus as a guide, this hints at a possible culture mismatch for women in OSS and that women participants would become attuned to these issues. Indeed, Prana et al. found that women place greater importance on the social aspects of participating in OSS projects relative to men [SE23]; according to their study, women value selecting a project that has friends and colleagues on it more so than men (64% of women vs. 25% of men), and more women find having a shared gender identity with fellow contributors to be important compared with men (37% of women vs. 1% of men). That is, women are more likely to seek social signals that indicate that a project will make room for them and others like themselves. Gender-related biases in male-gendered environments can also manifest in subtle ways through work practices and tools. For example, Mendez et al. [SE18] found that tools and infrastructure in OSS projects are implicated with causing gender biases. Their study showed that in 88% of the barriers, self-efficacy was identified; that is, for women newcomers with a lower sense of self-efficacy, tool issues can further erode their confidence. A follow-up study by Padala et al. [SE21] found that even when tool biases affect both men and women, women can be disproportionately affected: for example, a lack of support for cognitive diversity in OSS tools disadvantaged women significantly more than men.

(Organizational) Lack of women in top positions: A common problem in male-dominated industries is the concept of the “old boys’ network,” where men in power preferentially hand down opportunities to other men, to the exclusion of women. In the context of primary and secondary education, Peters illustrates how “seasoned professionals (typically White males) have sought to assist protégés who are ‘younger versions of themselves’ […] As women have entered school leadership, they have experienced limited access to productive mentoring relationships, further limiting their access to school leadership positions” [GL18]. The lack of women at the top can become a self-fulfilling prophecy, as men mentors in leadership may associate gender with performance potential and the women mentors are unavailable to help lift other women up [GL6, GL7, GL13, GL21]. Meanwhile, women who do succeed and become leaders often find themselves overburdened by mentorship relative to their male peers. For example, Corneille et al. note how senior women of color in higher education may feel “an obligation to serve to honor the legacy of those who provided mentorship throughout previous generations” [GL8]; similarly in sports Bower describes how “the demand for female mentors creates a problem due to the shortage of women in the upper levels of the organization. When there is a shortage of women at upper levels of management, women in these positions are overburdened with women needing mentors” [GL17] (see also [GL23]).

In OSS: In a study of over 700 OSS projects, Canedo et al. found that while 5.35% of contributors were women, women were relatively underrepresented in top positions on projects, making up only 2.30% of the total number of core developers [SE8] (cf., Wang et al.[SE30]). Of note, the authors reported finding no differences in the work activities of core developers along gender lines. According to Steinmacher et al., this extends mentorship for developers more generally: men and women are equally likely to perform mentorship in OSS projects [SE26]. So while men and women senior developers may perform mentorship at similar rates, there are relatively few senior women available to offer mentorship in OSS projects. It is unclear, however, whether this actually translates into greater workloads for women; Feng et al. found that women provided more mentorship than men, but the effect size was small [SE11].

(Organizational) Formal mentorship can fail to provide the support that women need: Even when organizations do recognize the value of mentorship and set up formal structures to encourage it, this does not guarantee that women will get the support that they need. In a study of mentorship of women in a medical residency program, McNamara et al. found that women’s relationships with assigned mentors tended to be transient and impersonal and that arbitrarily matching mentors to mentees did not go far enough in encouraging beneficial mentoring relationships [GL2]; in that study, the best mentor-mentee relationships grew organically from informal relationships. Researchers caution, however, that informal mentorship is not a panacea, as it tends to reproduce existing organizational and cultural barriers [GL10]. This also tracks with sources cited earlier showing how women tend to have less access to mentorship than men – having only informal mentorship channels may mean women receive no mentorship at all.

In OSS: Some OSS projects, particularly high-profile ones, have specialized programs with mentors’ selection criteria to identify mentors who are a good fit for the students [SE25], but formal mentorship practices are not widespread across projects. Literature suggests that OSS mentors are not usually formally trained and either voluntarily elect to mentor out of personal interest or are asked to do so by the community [SE4]. This could be a newcomer asking for help from a senior developer [SE10] or core members commenting on issues of newcomers, which is comparable to mentoring [SE24]. Relationships are then established between a more experienced developer and a newcomer after having had a couple of interchanges of email, leading up to exchanging personal messages on other platforms [SE9]. Kariri and Rodríguez [SE16] studied mentoring in the Stack Overflow community. Mentoring here also takes place informally; people help other people out by answering questions people post without being obliged to do so.

RQ2. What Strategies Can Help Overcome the Challenges Faced by Women Mentees in OSS Projects?

(Mentorship Practices) Women mentees benefit from women mentors: Women benefit from having mentors of all genders. That being said, there are particular benefits in matching women mentees with same-sex mentors. Leck et al. cite numerous other works that show how “compared to cross-gender relationships, single-gender dyads alleviate the difficulty of mirroring ‘male behaviours,’ increase interpersonal comfort, and provide more psychosocial and career development support” [GL7, GL16]; we note that Høigaard and Mathiseson ([GL9]) give a contrary result and did not find a significant difference between same-gender and cross-gender mentorships in their study in terms of psychosocial support or role modeling (n=36), which they attribute to all mentors in their study receiving training on mentorship and communication. Women lifting up other women may help break the cycle of women being underrepresented in OSS project leadership. Bosu and Sultana note that promoting and mentoring women to leadership positions as an effective solution to foster gender diversity in OSS [SE6].

(Mentorship Practices) Encourage having multiple mentors: Hybrid models of mentorship such as mentor networks, peer mentoring, and group mentoring can lessen the problem of not having enough women mentors to match to women mentees [GL13]. Moreover, Smith-Jentsch et al. remark that an advantage of online mentoring (as is the case for OSS) is that it is easier for mentees to enter relationships with a diverse network of mentors [GL1]. This diversity of mentors can span both personal backgrounds and skillsets; reporting on the needs of mentorship of black women in academia, Evans and Cokley explain how “a person of a different culture or sex may provide excellent research mentorship, but may not be able to sufficiently attend to African American women’s research interest or unique needs. For example, a primary mentor may be chosen for his or her research skills whereas a secondary mentor may be chosen to foster ethnic minority or sex/gender research and career interests” [GL5]. Likewise, for women contributors to OSS, having access to different mentors may help provide a stronger support network while also reducing the burden placed on individual mentors [SE12].

(Mentorship Practices) Make goals of mentorship (e.g., psychosocial support) explicit and anticipated outcomes: In general, having clear goals in mentorship helps ensure that the relationship is productive and impactful. For example, Trainer et al. found that task definition is an important design consideration in mentorship schemes [SE29]. Case in point, mentees working on user-facing, interdependent software develop both technical skills and social ties. According to the authors, it is important to structure mentees’ tasks so that can yield important learning and interpersonal benefits for the mentees; identifying what goals mentees have up front is key to matching them with tasks that advance those goals. Mentorship, however, goes beyond just skill development and career support but also encompasses psychosocial support, and this may be especially important for retention of women mentees [GL3, GL7]. Findings by Smith-Jentsch et al. indicate that, all things being equal, psychosocial support tends to be less frequent in e-mentoring contexts [GL1]; for this reason, Leck et al. argue that mentors in e-mentoring contexts should be explicitly reminded that adequate psychosocial support is a necessary function of mentorship [GL16]. OSS projects that want to encourage mentorship of women should be up-front about the purpose and intended outcomes of that mentorship and should think holistically about both career and psychosocial support aspects.

(Policy Instruments) Provide inclusivity-aware mentorship training: Mentorship programs should be geared toward inclusion through intentional training and guidance to raise awareness of the needs of women and minorities [GL3, GL5, GL19, GL20]. Studies indicate that having inclusive environments is significantly more important to women than men [GL7, GL15]. With the understanding that women may be less likely to seek mentorship (e.g., for fear of being seen as less capable), McNamara et al. argue that “[women mentees] should be educated about the importance of mentorship, given various strategies to initiate contact with potential mentors, and encouraged to actively seek and maintain their mentoring relationships” [GL2]. Likewise, as we mentioned earlier, gender-related differences in communication can create barriers for newcomers to OSS. Intentionally discussing these issues and challenging gendered humor [SE17] may promote a more inclusive environment. Brainstorming can also support minority team members’ engagement and satisfaction [SE2, SE13] by providing structure for all team members to voice ideas and by encouraging integration of every contribution.

(Policy Instruments) Monitor progress and allow women to exit dysfunctional mentor-mentee relationships: To minimize negative outcomes, mentoring programs should monitor the progress of mentees and enable them to exit from a mentor-mentee relationship that is not a good fit for them, and this may be especially important for online mentorship [GL1, GL7, GL16]; doing so also has the added benefit of creating opportunities to gather data and improve upon the mentorship program. Silva et al. [SE25] show that OSS projects can adopt mentoring coordination actions, such as monitoring mentors’ activities as a strategy to help the mentors with the challenges they face and reduce the odds of failure. While this can benefit all mentees during onboarding, it is significant for women contributors who may otherwise feel uncomfortable raising their concerns over a poor mentor-mentee relationship.

(Policy Instruments) Recognize and reward mentorship, especially mentorship of women: Mentorship requires a substantial investment of time and effort and should be rewarded accordingly [GL10]. Corneille et al. recommend highlighting mentorship success stories on institutional websites and that funding agencies should create incentives to encourage mentorship of women and minorities [GL8]. Other authors have suggested making space for mentorship by decreasing mentors’ workloads elsewhere, signaling that building mentor-mentee relationships is just as important as any other activity [GL5, GL23]. That is, mentorship needs to be treated as a first-class contribution to projects, and mentorship of women and minorities should be especially prioritized.

(Other Strategies) Connect women with online support communities: Joining an online community can be a means to overcome the challenges faced by female software developers in an OSS project; for example, FLOSScoach is a community that helps increase the self-efficiency of newcomers in an OSS project [SE27]. Another study by Abanoz shows that social support mechanisms can provide the necessary support for individuals with the activities they carry out and they create the necessary environment for individuals’ motivations [SE1]. In this way, they enable individuals to perceive themselves in a better situation and perform actions in line with possible selves, improving the low self-efficacy of women software developers that is mentioned as a challenge in OSS projects.

(Other Strategies) Create more equitable working environments for women: Related to gender bias in tools, the GenderMag method proposes a solution [SE7]; GenderMag, short for Gender Inclusiveness Magnifier, is a method for identifying gender inclusiveness barriers in software and generating ideas to fix those barriers. Padala et al. [SE21] used this technique with OSS teams to improve newcomer experiences in OSS projects.

Discussion

The goal of this study was to investigate the mentorship of women in OSS projects to support the onboarding process to prevent early disengagement of women in those projects. We observe that women in OSS face a variety of personal, relational, and organizational barriers to receiving mentorship. By situating the SE literature on this subject within a broader, cross-disciplinary context, we see these experiences are comparable to women professionals in other contexts.

By establishing that mentorship and women’s experiences of mentorship are comparable, this positions us to leverage theoretical insights and potential solutions from the broader literature. There are numerous approaches backed by evidence that can help ensure that women in OSS receive the mentorship they need, including adopting better mentorship practices, using policy instruments to promote a culture of mentorship excellence, and other strategies that increase women’s sense of belonging in the OSS community. This both reinforces the recommendations of SE researchers (e.g., matching women mentees with women mentors) and also highlights strategies that have been relatively understudied in our field (e.g., mentorship training programs and reward mechanisms for mentorship).

As noted by Stoeger et al., a key issue in the design of mentorship programs is a lack of awareness of the state-of-the-art within mentoring research [20]; we believe that acquainting SE practitioners with insights from the broader literature may be helpful. As for SE researchers, we believe that the wealth of scholarly resources on mentorship in other domains presents opportunities for replication studies and other forms of cross-disciplinary dialogue.

As with any mapping study, our work focuses on providing a broad overview of the subject matter rather than an exhaustive and in-depth treatment of the literature; there are many more studies available to us, including more recent works that were not in scope at the time we carried out our search on the SE literature. Moreover, as this was a cross-disciplinary review, we faced limitations due to a lack of institutional access to certain journals and proceedings outside of our domain. We do believe, however, that we achieved sufficient coverage of the relevant literature to be able to tell an accurate story of women’s experiences in mentorship. We also note that neither mentorship nor the experiences of women professionals are entirely the same across disciplines. The literature we drew upon comes to us from multiple, distinct traditions of mentorship research and practice. We do argue, however, that the findings we present from our two corpora are complementary even if they are not perfectly comparable.

Conclusion

Mentorship is a widespread professional practice across numerous domains for developing talent and fostering enduring communities of practice and one that deserves special attention in meeting the personal and professional needs of members of underrepresented groups. In this chapter, we presented initial findings from a systematic mapping study on mentoring women in OSS projects. Using sources both from within SE research on OSS and mentorship literature in other fields, we identified challenges experienced by women mentees as well as strategies that can be used to overcome these challenges. We hope that our work helps enrich the discussion on how mentorship can play an instrumental role in promoting diversity, equity, and inclusion in software development practice.

Acknowledgments

Sandia National Laboratories is a multimission laboratory managed and operated by the National Technology and Engineering Solutions of Sandia, LLC (NT-ESS), a wholly owned subsidiary of Honeywell International, Inc., for the US Department of Energy’s National Nuclear Security Administration under contract DE-NA-0003525. Images are used by permission. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of NT-ESS, the US Department of Energy, or the US Government. SAND2023-06679B.

This paper uses icons from the Noun Project by artists Adrien Coquet, Binpodo, Eucalyp, Numero Uno, and Stephen Plaster.

Table 20-1 Selection of sources from literature review on mentorship of women in the general (non-SE) literature

Table 20-2 Selection of sources from literature review on mentorship of women in OSS from the SE literature