1 Introduction

As qualitative researchers, we are often encouraged to think about the imbalance of power within any empirical endeavor (Raheim et al., 2016). Our positionality as privileged and trained scholars working in academia necessitates protecting our subjects from harm and risk (Arlinda, 2022; Fenge et al., 2019). In offline research this often means obtaining an informed consent and acknowledging participants’ vulnerability, while in online research this often means maintaining privacy and anonymization of data alongside ensuring strict data hygiene and the absolute lack of traceability (Yadlin, 2022; Yadlin-Segal et al., 2020). However, whether offline or online, over the years, risk and safety protocols have prioritized discussions about the participant, the subject in the study, or the studied community, overlooking some important aspects related to the risk and safety of the researcher within the changing landscape of academia and media as one (Fenge et al., 2019; Harries, 2022).

The valuable discussions that exist about the risk and safety of researchers mainly focus on sensitive topics and risk-prone interactions pertaining to the participant in the study (Barratt & Maddox, 2016; Dickson-Swift et al., 2008). Thus, the safety of the researcher is considered in the context of exposure to emotionally disturbing issues throughout the project’s data collection and analysis. This may arise when working with vulnerable communities with traumatic life experiences, research subjects with health-related issues, partaking in illegal activities, etc. (Bloor et al., 2010). Against the backdrop of this knowledge Kulnik et al. (2020) recently argued: “the management of [researchers’] unexpected vulnerability is often given insufficient space” (p. 3).

With specific emphasis on qualitative methods for online materials, growing scholarly attention is given to the impact of toxic traits in digital life on researchers [such as hate speech, microaggressions, “doxxing”, or extreme content; see: Conway (2021)]. These are only discussed in contexts of risk-prone interactions online (Arlinda, 2022; Barratt & Maddox, 2016; Conway, 2021; Mattheis & Kingdon, 2021) with little attention to the continuing risks post-study (i.e., once we conclude the study period). At the same time, very little attention is given to the holistic nature of digital research work, where studying online cases is always directly connected to offline existence of researchers. Whether in relation to our past experiences or through multi-sited movement during the period of study, digital research is always an endeavor that connects the online and the offline (Yadlin-Segal et al., 2020). In fact, it was only recently that formal general discussions and guidelines for online research work—such as the Association of Internet Researchers (AoIR) ethical guidelines 3.0 and the establishing of a thinking group on the topic—have started highlighting the need for protecting researchers alongside participants in online interactions.

Aiming to enrich this emerging discussion, we reflect in this piece on three methodological considerations relevant to risk and safety of researchers in both the research of day-to-day settings and topics (i.e., not risk-prone) and risk-prone settings. We ponder safety throughout and after the study period. Based on our own empirical research projects in the field of digital media studies, we focus on (1) Aya Yadlin’s reflection on the siloed nature of digital media studies, sexual harassment and the safety of women-researchers in ethnographic research spanning the online and offline, (2) Ruth Tsuria’s work on analyzing online discourse that triggers a researcher’s own complicated past and unrealized personal trauma, and (3) Asaf Nissenbaum’s reflection on qualitative scholarly immersion in the unregulated digital “frontier.”

The three cases seem far from each other in nature, aim, and relevant safety measures. Yet, the specificities of each situation discussed here, illuminate one of the most pressing concerns in digital research: There seems to be such range of issues to address in the sub-field of digital media studies that any endeavor to encapsulate the sub-discipline in its entirety in almost impossible. Hence, we present these different cases, from different online spaces, through different qualitative research methods, to highlight the need for protocols aimed at ensuring the safety of online studies holistically, as a wide umbrella that must cover online research as much as possible. The grave differences between the cases become a strength, a way to highlight the depth and breadth of our sub-field.

We focus on our own experiences in qualitative research of online contexts to stress the need for holistic protocols that inform, prepare, and assist scholars facing the risks and prices of qualitative studies in, of, and through internet-based platforms. Together, these help us highlight that qualitative researchers’ unique position of power is neither absolute nor consistent and is always an interaction between the online and the offline. It also helps us to highlight the importance of formats such as commentary and notes on journals in our field. The specificalities of digital research offer a unique opportunity to shed light on the conversation required regarding the safety of researchers in qualitative methods protocols in the digital society.

2 Sexual Harassment in Ethnographic Fieldwork Spanning the Online and the Offline

In 2016 I experienced sexual harassment during my own ethnographic fieldwork. I was a graduate student working on non-sexual related multi-sited digital ethnography, when a participant made gratuitous and multiple sexual and sexually-objectifying comments in one of our meetings. At face-value, I was doing everything “correctly” while journeying between online and offline settings. As a young woman working alone in the field, I made sure to minimize one-on-one interviewing in remote or secluded places, I prioritized group meetings where other female participants attended alongside male participants to avoid being the only woman present, and whenever possible I invited a friend or a family member to join me in fieldwork. But it only took one meeting alone with a key male informant in his home, a well-respected player in the community I was studying, to be sexually harassed.

The comments made by this informant had nothing to do with the topic of my research. It started with an unsolicited comment about my appearance and escalated throughout our meeting into sexual innuendo, him sharing graphic sexual preferences, and pressing me to share my own. It took me time to understand where the discussion was headed. I left in a rush, at a real loss for words and tools to make sense of the research experience. This event put an abrupt end to my data collection phase. I was unable to continue working on materials that concerned him, and as I was often told that “a good dissertation is a done dissertation,” I found ways to work through the heart-wrenching encounter, simply to put everything behind me. I have not conducted any fieldwork, offline or online, ever since.

Thirty years ago Green et al. (1993) highlighted the “general lack of attention given to issues of sexual harassment in field research” (p. 629). Not much has changed since, and we can see that the intersection of gender, ethnography, and safety is still neglected when it comes to women researchers conducting fieldwork (Harries, 2022). Academic personnel being predominantly male—an issue that becomes more pronounced with seniority—also frames the ways research roles, risks, and opportunities are perceived, taught, and regulated institutionally (Green et al., 1993). Thus, while sexualized interactions, objectification, and harassment during fieldwork are described by women researchers as a common experience, this continues to be marginalized in methodological discussions and training (Hanson & Richards, 2017; Harries, 2022; Kloß, 2017).

At face value, while this collaborative piece is about online methods, the harassment occurred offline. Yet, as I have been thinking about the situation for many years since, it seems that my preparedness for the situation, or lack thereof, had to do with the siloed learning experience of students researching online settings. Once committed to a digital media project, much of the training on risk and safety had to do with online settings. The topic was well addressed in training, yet the educational experience on offline settings was overshadowed by a focus on digital and online culture, research protocols, and the like. This siloed experience, while allowing professionalization experience in topics relevant to digital life, became a source for confusion and lack of knowledge as to safety and risks in research.

With regards to ethnography conducted in both the online and the offline Hine (2007) argues that the ethnographer:

[M]oves between online and offline as users of computer-mediated communication do… looking at the construction of boundaries and the ways in which different forms of communication are used to contextualize one another. It is open to embedding processes, looking both at the ways that lives are embedded into computer-mediated communication and processes through which computer-mediated communication is embedded into lives. (p. 617)

Contrary to what I expected, it was not anonymity or aggressive online behavior that jeopardized my safety during fieldwork (Boellstorff et al., 2012; Dibbell, 1993; Turkle, 1995). My own experiences in the field revealed that it is the “traditional,” longstanding offline-oriented fieldwork protocols that lag in terms of highlighting the possibility of sexualized reactions from participants in the field. It was the offline that caught me off guard. I had embarked on that portion of my data collection with much excitement, with an institutional review board (IRB) approval for my protocols stressing how I intend to protect my participants from risk, with a lengthy framing of my own power-position in the field, but with absolutely no understanding of systems in place for discussing the harassment I was about to face in the physical-offline field, not the digital one. My peers and I were able to express complex ideas about gender and intersectionality in the social sciences, about the safety of participants between the online and the offline, but had very minimal understanding of routes for reporting sexual harassment in fieldwork, and with no clear institutional channels for support, I decided to push through and keep it to myself.

Even as I type these lines, I keep asking myself: “But what if I am putting my informant (yes, the same one that sexually harassed me) at risk? Am I even allowed to reveal such information? Aren’t I supposed to protect him?” We are urged to protect our participants from harm. We are trained to prioritize their safety to the point where we, as women researchers, are not taught to protect ourselves. This lack of training means that the shared experience of sexual harassment in fieldwork is not reported and thus cannot be chronicled into effective institutionalized training, systems of support, and guidelines (Bloor et al., 2010). It means these experiences remain individual, isolated- both leaving survivors to deal with it on their own and making it an invisible phenomenon.

This, as other scholars have mentioned, is true in terms of studying sensitive and risk-prone settings (Chiswell & Wheeler, 2016; Green et al., 1993; Silverio et al., 2022), but as I have learned, becomes extremely important when studying the seemingly innocuous every-day context, moving between online and offline settings. To existing knowledge, we ask to add the important focus on this siloed experience of graduate students and early career scholars taking their first steps in empirical studies in the field. The “technical” portion of administering movement between sites, the different consent and safety protocols required for online vs. offline interactions, and the over-emphasis on me being the “powerful,” privileged scholar entering the field, eclipsed one simple fact: Harassment of women researchers in the field is not systematically approached and acknowledged in training and by organizational boards.

3 Secondary Trauma and Textual Analysis of Online Texts

During the process of completing the IRB application on my in-depth research of negotiation of gender and sexuality in online Jewish communities, like many researchers, I was asked to assess the risk to my participants. While I did not have participants per se, I thought deeply about how my research may cause harm to the community I was studying. I did not, however, consider my own risk or safety when conducting such research.

As I unpack below, I was researching the very same ideologies and concepts that harmed me as a young girl for many years, not fully realizing the trauma experienced over the years. In fact, it was only while working on a research project which was not explicitly related to my religious past (Jewish Orthodox) or specific community (National-Religious in Israel), that I have felt the impact of Secondary Trauma (Williamson et al., 2020). According to Williamson et al. (2020), Secondary Trauma (ST) is “the impact of indirect exposure to traumatic experiences” (p. 55)—that is, any engagement with or exposure to stimuli that relates to or reminds of a traumatic experience. While scholarship recognizes ST in various settings, less attention has been given to the impact of ST on researchers (Dickson-Swift et al., 2008, 2009). This is perhaps, as Williamson et al. (2020) consider, “due to traditional views of academic scientific endeavor as objective, detached and neutral, where researchers are not supposed to feel anything” (2020, p. 56).

However, recent academic discourse recognizes that in reality, researchers are not neutral, and are indeed impacted by their research. This is certainly true in cases where researchers work on sensitive and risk-prone topics, such as gender-based violence, as is the case in Williamson’s study. But it can also occur—as it did in my case—when researchers work in areas that are related to their own identity and culture but not sensitive per se. Here we must ask, what is the risk we take when researching areas we do not necessarily consider as potentially triggering for ST, and what safety procedures can be taken to minimize this risk?

Growing up as an Orthodox Jewish woman, many things were beyond my reach: Full religious participation, gender equality, and sexual education—to name a few. For many years, even after I ceased actively practicing the Jewish religion, I did not consider these impasses as traumatic, but rather as social discourses that were not aligned with my needs. But while researching ex-Evangelical discourse on Twitter (Tsuria, 2020) I encountered the term “religious abuse.” Members of this Twitter community discussed how they felt gas-lit by the religious ideology they grew up in, and how that negatively impacted their understanding of the world, spirituality, and their view of themselves. I realized I felt similarly, and for the first time recognized my own experiences as religious abuse. I felt hurt, and could not continue research into the Jewish Orthodox community. I also had to re-evaluate my other research areas, and stop some of my research projects that dealt with similar topics. I was experiencing ST caused by my own research.

Importantly, I asked myself how my trauma impacts my ability to research from a place of solidarity and care. As I have written in a separate publication, when qualitatively studying online texts, researchers should consider their capacity for solidarity with the people whose texts they are analyzing. I stressed: “When asking who owns the texts analyzed, we are also asking, whose voices does it represent, and do we stand in solidarity with these voices?” (Yadlin-Segal et al., 2020, p. 3) and pointed out that:

Digital texts are often written by living, contemporary users who did not volunteer their writings for research. […] a consideration of solidarity and ownership is needed when analyzing someone else’s words and potentially misrepresenting them, which can result in cultural, if not physical, harm. (Yadlin-Segal et al., 2020, p. 7)

But what happens when the texts analyzed are also a cause of hurt or trauma for the researcher herself? How can I stand in solidarity with voices that have caused me pain?

My solution to this was to utilize my own pain as a perspective for analyzing the texts: Not as the only or even most important voice, but as part of the discourse I was researching and presenting. I did not try to be neutral, but rather show the complexities of social norms in religious online communities. It was also, to a degree, helpful to examine this through the distance afforded by online texts (Turkle, 2017). In a way, perhaps it is the digital medium that allows for areas of ST to be researched in depth. But it is not enough to have the ‘safety’ of the screen—more thought needs to be invested in crafting protocols for dealing with digital material that is triggering.

Ultimately, the digital space did not protect me from the emotional impact of working with this material. Indeed, even from the distance of years and the safety of the screen, my trauma resurfaced and hindered my ability to dig deeper into the areas of digital media in which religious abuse is discussed. It is only through this collaborative piece that I am able to process and publicly address this risk.

4 Qualitative Study of an Unregulated Digital Frontier and the Risks of a Methodical View

My first full-scale research project, which led to my first publication (Nissenbaum & Shifman, 2017), was based on a then-relatively unknown website—4chan.org, and specifically “/b/”, its Random channel (or sub-forum). My interest in digital community and identity led me to focus on this unique community, which, despite its ephemeral and anonymous platform, became a main hub of digital culture and meme creation (Bernstein et al., 2021; Milner, 2016).

I had frequented this community as an observer for some time before commencing my research, and thought I knew what was involved in delving into this notoriously unregulated space. Whatever public reputation it had was earned through nihilistic and incessant trolling of various web forums, eventually being dubbed the “internet hate machine” (Coleman, 2014; Phillips, 2015). As such, this site of study, in comparison to the above cases, can be understood a-priori as a risk-prone context. Accordingly, the discourse within the site was vulgar, acidic, and openly hateful towards most social groups, especially ethnic minorities and women. However, my aim was not to criticize the site’s discourse (as its faults were clearly observable) but to investigate how its sense of community is formed and maintained. Thus, I needed to adhere to qualitative research principles—accept, to the possible extent, my subjects’ worldview and try to see through their eyes (Geertz, 1985; Markham & Baym, 2009).

This task was more challenging than I had anticipated due to the nature of a research-oriented view. The combination of an unregulated online forum and a community with a propensity for trolling meant the forum was teeming with extreme visual content, including many disturbing violent and/or sexual images. While these were evident on any casual visit to this site, the option of looking away and disregarding them was relatively easy. This was not the case when conducting the research itself—rigorous studying of this community demanded methodic selection of the content being analyzed, i.e., specific discussion threads where community boundary work was evident.

Yet, the main risk factor in this project arose well after its conclusion. This was when the study was accepted for publication—despite the obvious positive implications, having it publicly available in a prominent journal was also a cause for concern. As noted above, the 4chan community was known for pointing their aggression toward ideological opponents, those criticizing the site, or shining a spotlight on it (Colley & Moore, 2022; Phillips, 2013, and both text and comments to this note by; boyd, 2010). Moreover, the site clearly included many members with the skills necessary for infiltrating personal accounts and data of an average web user (Coleman, 2014). Thus, gaining recognition for this research work could put me, my co-author, and others at the focus of a digital harassment campaign or other harm by a group with a wealth of experience in such behavior.

Turning to the implications of this experience, I must note the difference between myself and my fellow authors in this collaborative piece, and specifically how it is mediated by positionality (Arlinda, 2022; Fenge et al., 2019). The challenges involved in this study had a profound effect on me. Alongside constantly recalling some of the disturbing images I encountered while conducting this study, I also continually look over my “virtual shoulder” with concern that this text may finally draw the 4chan community’s attention toward me, as the site has only become more extreme and political since my study (Tuters & Hagen, 2020). That said, unlike the other accounts in this paper, I remain an observer of the risks and hurtfulness associated with studying 4chan, rather than being a direct or likely subject. Being male, passing as White and belonging to my country’s ethnic majority, and being from a developed country, I was mostly aligned with the site’s imagined community, and that of digital culture at large (Milner, 2012; Nakamura, 2013). Thus, while for many others, the violence, sexism, and racism in the study’s corpus would have been directed at their own groups or identity, my position allowed for a removed view that was somewhat shielded from direct implications or hurtfulness.

Looking back, my main takeaway is that studying unregulated digital spaces should be thought of as a specific and noteworthy case when considering the safety of a given analysis method for those practicing it, not only for sensitive topics, but also in relation to regulation of researchers’ content safety and post-publication personal safety. This is especially true when observing what may be thought of as internet frontiers—places where few researchers have engaged with in a systematic way and for which very few research regulation protocols have been established. Such systematic studies involve a markedly more detailed and focused experience than that of a casual visitor, and that needs to be taken into account. Finally, positionality should be a key consideration across both these issues. As members of disenfranchised groups are likely to be more vulnerable in such cases, it is both the responsibility and interest of academic institutions to provide the support needed to enable diverse research, especially for frontier, unregulated spaces.

5 Conclusions

Our aim in this collaborative piece was to enrich the emerging discussion on researchers’ safety and risk in qualitative research of online contexts. Barratt and Maddox (2016) argue in this context that “the safety of researchers working in digital spaces needs to be properly considered and safeguarded with the same care as is applied to conventional research engagements” (p. 12). The three cases reviewed here allow us to add that such safety must be (1) understood as an intersection, (2) protected to the same extent as participating communities’ safety, (3) acknowledged and highlighted from early stages of research training and professionalization. These insights and recommendations are elaborated below.

  1. (1)

    Safety as an intersection: As delineated through three purposely distinct cases, different in their approach, studied space, and methodology, we stress that protocols should look at researchers’ safety in online research as an intersection of platforms, “seasoned”/conventional spaces and the frontier, the online and the offline, tools, approaches, and methods throughout and after a study. Thus, protocols should offer a wide-as-possible umbrella to deal with risk issues not just in cases of studying risk-prone contexts or simply as a legislative issue. Protocols should offer go-to-instructions both before and after conducting the study vis-a-vis ensuring that enough resources are indeed in place to protect and aid researchers’ safety. While researchers can independently implement practices of self-care and stepping away from data material (or even the screen), these are not enough on their own, and should be accompanied by institutional systems of support that are not silo-dependent.

  2. (2)

    Safety of researchers as safety of participants: Noting that qualitative research safety protocols often prioritize participants and subjects’ safety, we suggest that more should be done to ensure parallel, similar measures for protecting researchers from harm. It should be acknowledged that research comes with its own unique emotional and mental price, derived from the committed, deep, and empathic gaze needed for qualitative research. As shown in the cases above, in going beyond casual viewing and/or participation, researchers intimately engage with communities and participants in ways that often have impactful, lasting consequences. In this context we specifically stress that protocols should ensure measures for protecting researchers’ safety even when the topic of study or the data collected online are not a-priori associated with harmful acts, practices, communities, or information.

  3. (3)

    Early start on training: Finally, we wish to stress that education on these aspects is key. Researchers conducting qualitative studies should be made aware of the personal emotional challenges that may lie ahead. Accordingly, protocols on researchers’ vulnerability should be introduced as early as possible in research methods classes, professionalization workshops, research assistants’ training, and even academic networks (such as academic X (formerly Twitter), for example). This should be true for personal data storage, working with past experiences, involvement in the field, etc. If we take seriously the idea that institutions should put forward efforts into establishing these protocols and routes of support, then we, as an invested community, must constantly stress how and when to approach these venues and take part in advocating their necessity.

One final note is on the importance of formats such as this “brief communication” collaborative piece. Writing about trauma, risk, and safety is not always a process that can be published as a full-blown empirical research manuscript or as full guidelines for research projects. If one of the main problems in this context is the lack of proper documentation of harm and risks, then testimonies shared via commentary and notes on journals related to the topic become an important tool for awareness. When collected across our sub-field they become an important starting point for solutions. A format such as this essay allows us to bring together cases that seem distinct but can help shed light on each other specifically through differences (online-offline dichotomies, risk-prone and not risk-prone, and the like). Gathering, documenting, exploring, and grounding these cases together in existing literature thus becomes a way to overcome the lack of institutional solutions to these challenges.