There are many studies that investigate the experiences of marginalized software developers; however, they tend to include gaps in how people from a marginalized background compare against a majority group. Many researchers focus on where historically marginalized groups do not measure up, missing the opportunity to understand where participants are in fact excelling. Likewise, asking research questions that only seek to surface deficits rather than successes can unintentionally create an unmatched and negative precedence that participants do not hold about their experiences.

In an effort to set a new precedent for conducting and reporting research, we build on asset-based design to propose what we call abundance-based reporting. In this chapter, we outline how to investigate, report, and build interventions with historically marginalized communities.

Introduction

The focus on diversity, equity, and inclusion in software engineering is less often getting brushed off as a side project, becoming a core and pivotal part of moving the field of software engineering forward [1]. As this awakening continues, empirical researchers may be seeking guidance on how to appropriately investigate and to continue empowering developers in the margins. While many researchers are realizing the value in doing this, they are struggling with the practical approaches and ethical considerations they should take to do this. At the same time, there are also researchers who are trying to determine why it is important to center historically marginalized perspectives. The briefest answer we can offer to that group of researchers is that centering the margins provides us with great insight on how to successfully prepare for a multidimensional workforce in the future. One common reason some researchers fail to center the experiences of those in the margins is because it may not seem practical for them to do. Another reason is likely because they don’t see how it can impact their research long-term. An antithetical point that many researchers with this perspective often fail to recognize is that they are doing their own research a disservice when they exclude marginalized perspectives. There are unique experiences that live in the margins that can help us understand bespoke solutions to better serve the masses.

Throughout this chapter we will give examples and outline strategies for how researchers can overcome the issue of practicality and explain the longitudinal value of engaging deeper with people from historically marginalized communities. These strategies include approaches to initiate contact with organizations that cater to specific marginalized groups, approaches to engage with participants from study recruitment through data collection, how to report and share findings about their experiences, and finally ways to build a sustainable long-term relationship that can be mutually beneficial. Finally as we share these strategies, we also include exemplars of how this has been done successfully. Doing so gives us a model to follow and to adapt to our specific research settings. Many of the examples we present come from fields outside of traditional software engineering research; however, they are just as valuable and applicable to the work being conducted in our research community. That said, we encourage readers to consider this chapter to be a guidebook to researchers. Specifically, we hope this helps them feel encouraged to expand their view of how alternative methodologies can be relevant for their research and feel empowered to apply them in their work.

How to Center Marginalized Perspectives in Studies

Before understanding how to best center historically marginalized perspectives in research studies, let us first define what we mean by historically marginalized. We draw on the following description of marginalized groups [12]:

Historically marginalized communities

are groups who have been relegated to the lower or peripheral edge of society. Many groups were (and some continue to be) denied full participation in mainstream cultural, social, political, and economic activities. Marginalized communities can include people of color, women, LGBTQ+, low-income individuals, prisoners, the disabled, senior citizens, and many more.

A few of the marginalized groups we will be referring to in this chapter are across physical abilities and race. However, as many empirical software engineering studies have not yet investigated the full range of marginalized developers’ perspectives in depth, we should draw on other adjacent contexts and fields of science to glean insights on how to study and report on marginalized developer experiences successfully.

A few approaches some have taken to do this well are as follows:

  • Assets-based design: Asset-based design is a unique approach that leverages the strengths, existing knowledge, and institutional resources that a group may already have as core, thus building research and tools centered in that [14, 16]. In these settings, these assets are intentionally being used from a non-deficit perspective.

  • Joy-centric and celebratory perspectives: Joy-centric work strategically takes a non-deficit perspective to working with marginalized groups by focusing research on what the group or community celebrates [5]. This perspective of engaging and conducting research mirrors what a group chooses to celebrate – even if that may not mirror the research framing a researcher’s community may choose to highlight. You can consider this approach to be intentionally “in spite of” everything that may be negative and “easier” to build on based upon prior research.

Figure 29-1
An illustration denotes a set of 10 snippets of different pages arranged in 2 rows. The pages denote titles, texts, layouts, and infographics.

Speculative design workbook used in [6]. This can be used as a reference for future studies.

Full size image

  • Design fiction and speculative design: Design fiction is a specific approach to speculative design – a helpful research methodology that allows study participants to be critical of a design of a system and creatively imagine a new one [2]. In this approach participants are asked to describe what an ideal system for them could look like. This exercise allows participants to be in the driver seat of designing the tools and often results in supplementary design artifacts [6, 13]. (See Figure 29-1 as a reference.)

From these works, we can learn a lot about what it takes to be successful in highlighting perspectives the way that participants would like to be recognized. We should also note that each of these approaches is rooted in collaborating with the community at various stages of the research process. In the next section, we will highlight specific approaches within these paradigms that software engineering researchers can learn from.

Putting Methodologies in Action

In highlighting how researchers have investigated, designed, and built for joy in order to support the marginalized developer and technologist experience, we learn about the variety of formats that have been applied. Now that we have outlined specific approaches on how to center marginalized perspectives, we want to be concrete about approaches through which researchers have applied them. In the next subsections, we highlight two case studies for two different types of marginalized groups: one investigating the experiences of blind and low-vision (BLV) software developers and the other of Black and African American technologists.

Case Study: Investigating Online Communities for Blind Software Developers

To better understand how blind and low-vision (BLV) software developers use online communities as a resource for their needs, Johnson et al. investigated the use of Program-L [9]. Program-L is an online community “for users of Access Technology involved in programming to discuss any technical problems which are related to either the hardware or software they are using.”Footnote 1 In this particular study, researchers wanted to uncover the variety of help-seeking behaviors novice developers engaged in when participating in a community specific to two of their identities. One identity is their professional identity as a software developer, and the second is a personal identity as a person who is blind or has low vision. From their analysis of four years of novice behavior, authors were able to build a taxonomy of novice types and recommend design interventions for future demographic-specific online programming communities.

From this study there are several approaches taken that we highlight. One is the approach taken at the beginning of the study to contact the owners of the Program-L mailing lists before conducting the study. In this email (shown in Figure 29-2), the research team took several steps to establish legitimacy before conducting the study. For instance, the research team identified themselves and the work they had done previously. This showed to the community owner they had conducted similar studies in the past that the community owner could check. Next, the research team outlined their intent with analyzing the community and clarified that it would be conducted under the guidance of a research review board. Finally, the research team also made themselves available to have a one-on-one conversation with the community owner so that they might answer any questions or address any concerns the community owner might have.

Figure 29-2
A set of lines of text represents the format of an e-mail for a discussion on overlapping interests. It comprises 4 paragraphs.

Initial authorization email used in [9]

Full size image

There are several approaches to conducting research that center marginalized developers, for instance, asking permission before conducting studies, even if it may be a resource that a member of the research team may already belong to. It may not always be clear whom the permission should be requested from – in that case we would encourage researchers to reach out to members of the organization to figure out who the authorized leader in the community may be. Likewise, engaging in conversations with leaders and sharing the research findings back to the organization is one way to make sure the work conducted has the impacts intended.

Case Study: Technology for Black Lives Project

As another example of how researchers have been able to practically center historically marginalized experiences, we reference the ongoing Technology for Black Lives study [3]. In this project the research team is investigating how Black software developers and technologists are using, curating, and creating resources and tools in support of the Black lived experience. Authors draw on the archival narrative styles of scholars such as McIlwain [11] to investigate what technology’s role is in supporting the Black lived experience and what it means to design for it.

Although this study is not yet completed, the researchers have been publicly vocal about their study – publishing articles about the work as they collect data [7]. The research team has created a website (as shown in Figure 29-3) to announce the study. On this website, the research team publicly shared a variety of information pertaining to the study. Materials available on the public website include the recruitment flyer describing eligibility as well as compensation, who members of the research team are, a brief description of the projects, the intended project goals, how to contact the research team, and supplementary readings and resources.

Figure 29-3
A screenshot of a web page denotes headings, descriptions, and a menu bar at the top. The title reads Technologists for Black Lives. It indicates the details of eligibility, time commitment, compensation, and principal investigators.

Website for the Technology for Black Lives project. The website includes project descriptions, eligibility, time commitment, compensation, and principal investigators. There are also links to find out more about the research team, project goals, contact information, and supplementary resources

Full size image

From this project, there are several successful approaches to highlight. One is the fact that the research team created a website to share during the recruitment phase of the project. Having a public presence attached to the work can help increase trust from participants about how the study is being conducted, accountability on the researchers’ part, and transparency for potential participants and external researchers interested in tracking the work. Another attribute of this project to highlight is how researchers have used snowball sampling to recruit participants. Working with historically marginalized populations with a specific skill often requires researchers conducting studies with a network of participants that are likely to engage in some overlapping communities. In these settings, respectfully engaging with the network requires care. Researchers from this study used snowball sampling to recruit participants when applicable, which often resulted in participants referring to their personal network of technical colleagues – which is of great importance to them. Although the research team makes several resources public, they do not list all participants that have participated in the study without their consent. This type of intentional sharing has been helpful in making sure that the researchers have kept participants protected and that the research team continues to respectfully conduct their research.

The Dos and Don’ts of Reporting Marginalized Experiences

Recruiting Historically Marginalized Populations

Critical to the ability of conducting research that recognizes and considers marginalized populations is the ability to reasonably recruit research participants from these historically marginalized groups. In many cases, researchers may not already have relationships in these communities (and that’s okay). But how does one initiate a collaboration to form research relationships with a specific group?

DO make explicit effort to recruit participants from historically marginalized populations by tapping into your existing networks and relationships.

In this case, you want to focus on individuals and organizations with connections to the target audience or population. The goal here is to build trust through a trusted connection – individuals from historically marginalized communities may be more willing to engage in research if they trust the person conducting or connecting them to the research [4].

DON’T bombard individuals or organizations with requests to participate in your research.

There are many reasons you may not be getting responses when initially reaching out. If you’re contacting someone in industry, their lack of response may be about timing and their finding the opportunity to think about the opportunity and/or respond. If you’re contacting individuals or organizations that are community-centric (e.g., do work that services the community, not a company), their time is not only valuable but likely overloaded with existing commitments.

WHY? By tapping into existing trusted mutual networks and connections, you are building trust by association, which increases the likelihood for response, engagement, and building meaningful (and sustainable) relationships [4]. This essentially gives you a “foot in the door” to begin to build your own network. When reaching out, impatience and constant attempts to make contact can be off-putting and deter the response they originally intended to send (or further interactions in general).

Collecting Data from Marginalized Populations

Once you’ve acquired connections with your target audience for recruitment, the next obvious step is planning for and then conducting the data collection process. So the next question is, how do we build and maintain trust so that we can collect rich data that accurately reflects participants’ lived experiences?

DO be transparent about consent for participation and how the data will be collected and used.

When engaging any group in research, but especially historically marginalized populations, one should always begin and proceed with a mutual understanding of the procedures involved [4]. This includes being explicit about what the group’s participation entails, what data will and will not be collected, and how the data will be used in the near and distant future.

DON’T change research plans or directions without informing participants.

Research is fluid in that plans for collection or analysis may shift. Likewise, as research plans change and data be used in alternative ways than previously outlined, researchers should notify participants from which you have or plan to collect data and receive an updated consent.

WHY? The goal of any research collaboration should be to build trust for sustainable collaborations and outcomes. When the research is not conducted in a transparent and inclusive manner, this runs the risk of breaking trust [4] and has potential ethical implications [Chapter 9, “The Role of Ethics in Engineering Fair AI (and Beyond)”]. Even when not collecting data directly (e.g., via interviews), transparency is key to ensuring those whose data you are collecting and analyzing do not feel manipulated and used, rather than seen and heard [10].

Reporting Insights from Marginalized Populations

After data collection and analysis are complete, researchers are tasked with the sometimes daunting task of consolidating and reporting their findings. While it is always important to be mindful of how we report research findings, it is especially important to take care when reporting insights from historically marginalized groups. So the next question is, how do we accurately and respectfully report insights from these populations?

DO be explicit and factual about the demographics of the populations in your sample.

We should only report insights that clearly link to the questions we’ve asked or the data we’ve collected. It is also okay to make broad classifications or assumptions based on the given demographics. For example, if a participant states they are from Nigeria, we can assume they may belong to a broader African culture.

DON’T make narrow assumptions or claims that you did not collect data to support.

If a clear, and fact-based, connection cannot be made between insights and the data collected, it is likely that the assumption is too narrow. For example, while one can reasonably assume broader culture from geographic location, researchers cannot assume factors such as socioeconomic status from this information alone.

WHY? The assumptions researchers make may not always be true. In fact, it may be reinforcing stereotypes of a group and propagating false information. Furthermore, when done incorrectly, this can build false foundations for future research efforts.

Stating Your Positionality

In conducting studies on historically marginalized populations, the research team may or may not identify with the same background as participants. Either way, it is helpful for the audience consuming that work to be aware of the context the researchers implicitly bring to their analysis of the data.

DO acknowledge your positionality via a statement that clarifies your background in relationship to the specific demographic the research team is studying.

This can be done in a “Positionality Statement” or “Researcher Self-Disclosure Statement” in the Methodology or Introduction of a paper (See “Researcher Positionality Statement” in [8] for reference).

DON’T impose your positionality or inject your experiences upon participants while conducting the study or analyzing the data.

Researchers can resolve this by taking additional data validation steps. This could include having another researcher analyze data who is familiar with conducting studies where they have a different background from participants and comparing findings.

WHY? Researchers, especially when conducting qualitative analyses, are trained to identify various confounding factors (e.g., background of participants) to better understand how to analyze and report their findings. Likewise, it is also important for researchers to consider the effects their own identity and experiences might have on that process. This becomes very insightful when researchers choose to propagate and replicate the work of others.

Sustaining Relationships with Communities

Once relationships with historically marginalized groups or the organizations that serve them are established, it is critical to find ways to sustain them. Likewise, it’s important that this is done in a way that helps manage both the researchers’ and participants’ energy wisely.

DO provide multiple suggestions to participants and community organizations on the variety of ways to remain connected.

In doing so, researchers should make sure they are considering the size and goals of the participants or organization they are working with. For instance, every participant may not be interested in being featured at developer-specific conferencesFootnote 2 such as Strange LoopFootnote 3 or NeverWorkInTheory; there may be more of an interest in events like AfroTech,Footnote 4 Grace Hopper,Footnote 5 or even alternative forms of dissemination such as museum exhibitions that they would rather be featured in. In summary, researchers should be open to what may be most meaningful to participants and be sure to welcome their perspectives.

DON’T hide findings or final reporting from participants.

They should be aware of how their data was used as well as insights gathered from their experiences shared. It is understood that there may be a delay in responding and sharing findings back to participants, but this should be done in a reasonable time frame.

WHY? It is important to find strategies that make the relationship mutually beneficial. This helps the relationship flourish beyond the initial timeline of the project. Continuing the relationship should also be more than just having them participate in future research studies, but also supporting their initiatives that may be outside of traditional research conferences (e.g., community-based workshops) [15]. Keeping this in mind will help build a longitudinal relationship that can reinforce trust for their connections with the broader research community as well.

Conclusion

In closing, we encourage readers to continue to build on previous studies investigating approaches to supporting marginalized developer experiences. This especially includes conducting studies about how communities have been able to empower themselves. We hope that this work can serve as a guidebook for those understanding the best way to empower historically marginalized software developers and simultaneously contribute to the empirical fields of human aspects of software engineering.

Key Takeaways

  • Do be mindful of how to engage with, conduct research with, and report on marginalized experiences.

  • Do use case studies as a template for success and to avoid pitfalls in the future.

  • When centering marginalized experiences, try to intentionally consider approaches that center the assets of a community, what attributes they celebrate, and how to respectfully consider their perspectives.