GangstaLife: Fusing Urban Ethnography with Netnography in Gang Studies

Recent research on street-involved populations has documented their online presence and has highlighted the effects of their online presentations on their lives in the real world. Given the increasing conflation between the online and offline world, contemporary urban ethnographers should pay increased attention to their participants’ online presence and interactions. However, methodological training of this sort is still in its infancy stages and has not yet evolved to guide the growing number of researchers undertaking this form of research. This article draws from our experiences using social media in our urban ethnographies with criminally involved groups, to examine the benefits, risks, and challenges of drawing on social media in urban ethnography. It is intended to serve as a foundational piece that will hopefully ignite scholarly dialogue, debate, and methodological training relating to deploying social media in urban—and specifically—gang ethnography.


Introduction
With the narrowing of the digital divide, many disadvantaged populations across the world have access to the internet and social media (DiMaggio et al. 2004). Consequently, many of the groups that urban ethnographers often study are active social media users, including those who are socially and economically marginalized, people experiencing homelessness, at-risk The lack of methodological guidance on this topic is likely affected by generational differences between academic supervisors and students. Many of today's university students and junior scholars are "digital natives" 1 -they have grown up in a world characterized by the normalcy, pervasiveness, and near-necessity of computers, cellphones, and social media (Prensky 2001), though ranges in digital skills vary. In contrast, the majority of graduate supervisors and senior mentors are not "digital natives." This disconnect raises important concerns in terms of graduate training and mentorship, especially since the increasing recognition that urban ethnographers should strive to bolster being "there" with being "online" may result in more urban ethnographers deploying social media in their own research despite limited methodological guidance. As a result, those who may inadvertently find themselves fusing traditional neighbourhood ethnographies with digital research (such as Urbanik) or those who consciously decide upon this approach (such as Roks), are left to adopt a learn-asyou-go strategy. Considering that urban ethnography can pose notable risks for participants and researchers, the learn-as-you go method may further exacerbate these harms. Hence, it is imperative that urban ethnographers initiate serious methodological discussions about this novel approach to ethnography, as it raises new and important questions about how we should study street-involved populations.
In this article, we flesh out some of our personal deliberations and challenges relating to fusing urban ethnography with studying our participants' online worlds. We draw from our respective ethnographic studies with gang-involved participants, one based in Toronto's Regent Park neighbourhood (Canada), and the other in The Hague's "Forgotten Village" (the Netherlands), in which we separately concluded that we could not neglect our participants' social media experiences; their social media accounts and interactions were a goldmine of information for our ethnographic findings (Roks 2017;Urbanik and Haggerty 2018). This article proceeds as follows: We first outline our research projects and describe how and why we decided to incorporate social media into our respective ethnographies. Here, we present two distinct approaches of incorporating social media into ethnographic research, which build upon on Gold's (1958) four archetypical fieldwork roles, and discuss their respective advantages and disadvantages. Second, we describe how we made use of these online worlds, what kinds of insights these approaches yielded, and aspects we struggled with. Third, we highlight some practical and ethical considerations and potential pitfalls that may arise whilst conducting netnography. Our goal is twofold: to highlight the possibilities and dangers of this new research tool and to ignite greater scholarly debate on the subject.

(Un)Intentionally Initiating Netnography
First Fieldsite: Regent Park, Toronto (Canada) Urbanik's study is based in Toronto's Regent Park neighbourhood. Prior to its ongoing revitalization, Regent Park was Canada's oldest and largest social housing project, with all 69 acres dedicated to social housing (TCHC 2016). In 2006-at the onset of the revitalization-unemployment rates more than doubled the city's average, and almost 68% of households subsisted below the low-income cut-off (Horak 2010, 7). More than 1700 youth lived in the neighbourhood, 57% of residents were under 24 years old, and 37% of households were single parent (TCHC 2007). In addition, 78% of residents were foreign born and almost 80% of residents identified as "visible minorities" (Horak 2010, 6). The neighbourhood had Toronto's highest homicide rate from at least 1988 to 2003 (Thompson 2009), which undoubtedly contributed to its stigmatization as an alleged "haven for single mothers, welfare families and deviants…a magnet for crime and drug problems" (Purdy 2005, 531). Despite hopes that the revitalization's mixed-income model would hamper neighbourhood crime and violence, news media maintain "gangs, drugs and guns still rule Regent Park" (Warmington 2013).
Between 2013 and 2018, Urbanik spent over 12 months "deep hanging out" (Geertz 1998) and conducting open-focused interviews (Lamont and Swidler 2014) with neighbourhood residents-and specifically, gang-involved men. Although her core participants changed from year to year, she spent most of her time with a group of approximately 23 men (16-47 years old), predominantly of Caribbean and Somali backgrounds. Several of these men were gangaffiliated, many were involved in serious violence, including assault, robbery, and firearm offences, and almost all were involved in drug trafficking (marijuana, crack, powder cocaine, Oxytocin, Percocet, heroin, and crystal meth) (Urbanik 2018). As is common for many disadvantaged and racialized young men in North America, several participants were trying to establish (and some, successfully established) themselves in Toronto's rap scene, as they considered rap as one of their only viable career choices (Sköld and Rehn 2007). Consequently, many of them relied upon social media platforms to share and promote their rap music, music videos, upcoming shows, and to enhance their overall "street cred" as rappers and gangsters (Urbanik and Haggerty 2018). During the course of the study, several young neighbourhood men were shot and killed, including some of Urbanik's participants, which many residents attributed to rap rivalries or longstanding inter-neighbourhood "beefs." Since Urbanik was also connected with many of her participants on social media platforms such as Instagram and Facebook, she was able to witness the evolution of several such disputes, both in person and online.
Second Fieldsite: The Forgotten Village, the Hague (the Netherlands) Roks's study was situated in a small neighbourhood in The Hague, Netherlands, colloquially called the "Forgotten Village." Home to about 1400 residents, The Forgotten Village is an impoverished and highly racialized area that experiences many challenges. In the 1980s, it gained particular notoriety as the Dutch Crips claimed it as their home base, which they refer to as their "h200d" (Roks 2017). In 2015, almost 73% of neighbourhood homes were social housing, residential mobility almost reached 37%, and over 73% of residents were of non-Dutch origin, mostly of Surinamese, Moroccan, Antillean, or Aruban descent (Municipality The Hague 2015, 7-8). Forty percent of residents reported feeling unsafe and identified problematic youth groups, and one active youth group in particular, as the area's biggest problems. In 2011, local police identified that of the city's 385 known "criminal youth group" members, 69 resided in the Forgotten Village. Neighbourhood residents also struggled with mounting vulnerability and poverty, a higher prevalence of health concerns, staggering school dropout rates, and mounting tensions between residents (Roks 2017).
For three years, between 2011 and 2013, Roks conducted ethnographic fieldwork in the Forgotten Village. His study included semi-structured interviews, informal conversations, and ethnographic observation with 150 participants, including neighbourhood residents, social workers, police officials, as well as gang members. The crux of the project centered upon the lives and criminal careers of 60 participants (18-65 years old), predominantly of Surinamese and Antillean backgrounds, who were current and former members of the Dutch Rollin 200 Crips. At the onset of the research, the Dutch Rollin 200 Crips consisted of some 50 members (15-40 years old), predominantly of Surinamese background. The group engaged in serious criminality, including violence and drug trafficking (marijuana and powder cocaine), with several members being incarcerated during fieldwork for their role in stabbings, assaulting police officers, weapons charges, and possession of illegal narcotics. To further supplement his data, Roks monitored and analyzed the social media activity of some of these men and 40 street-oriented youth, utilizing platforms such as Instagram, Facebook, and Twitter (Roks 2017).

Field Roles in the Digital Era
According to Katz (2019, 17), ethnographic research refers to "the creation of primary data through personal interaction with research subjects in a project of creating or refining generalizations about human social life." Traditionally, ethnographic immersion within a space or group required physically "being there" (Geertz 1998). However, the 2000's ushered in greater fluidity in the object(s) of ethnographic inquiry (Wittel 2000), including a growth in multi-sited fieldwork as compared to single-sited spatially defined localities (Hannerz 2003). The urban ethnography of today marks yet another transformation-urban ethnographers increasingly spending time in, and studying, virtual spaces in tandem with physical streets (Lane 2015;Stuart 2019). Undeniably, the virtualization of everyday life has blurred the boundaries between the online and the offline (Leander and McKim 2003, 223), thereby complicating and challenging notions of "place, locality, and identity" (Ilan 2015, 72-73). Hence, we argue that the rise of netnography has significantly altered ethnography's long-held emphasis on "being there," and relatedly, our roles as researchers, and the nature of our interactions with research participants.
The relative novelty of incorporating netnography into urban ethnography raises important questions about the roles that ethnographers assume in these empirical endeavors. Traditionally, research roles in sociological field observation broadly fall into one of these categories: (1) complete participant; (2) participant-as-observer; (3) observer-as-participant; and (4) complete observer (Gold 1958). These roles are not fixed; they can change depending on the nature of the research project and stage of fieldwork, and usually shift from being more observationto participation-based as the study progresses and research relationships strengthen (Adler and Adler 1987). Although the online world and social media in particular have certainly transformed many facets of human interaction, we argue that the virtualization of everyday life does not require an entirely new methodological toolkit. Instead, the incorporation of netnography into urban ethnography makes space for research roles that are not currently fleshed out in existing literature. Hence, we argue for the recognition of two additional field roles on either side of the participant-observer continuum: the One-Way Mirror Approach (observant side) and the Glass Window Approach (participant side).

Watching from the Shadows: The One-Way Mirror Approach
The One-Way Mirror Approach entails a one-way study of our participants. In this sense, social media platforms serve as a tool that enables researchers to "see" participants' lives without participants knowing they are being watched or being able to reciprocate. Roks utilized this approach during the first period of his ethnographic research. In January 2011, his primary research contact from an earlier study-the gang's founder and leader, Raymond 2 -was incarcerated (Roks and Densley 2019). Despite this hiccup, Roks was eager to locate a field site where he could initiate learning about The Forgotten Village's social dynamics. With a youth worker's approval, he began spending time at a local community center where approximately 50 neighbourhood youth (12-20 years old) would hang out, several times a week for six months. Despite his efforts to build rapport with youth who frequented the center, they feared that he was a "snitch" [police informant], or the "po-po" [police]. Although the youth gradually became accustomed to Roks's presence, they were still unwilling to interact with him, thereby limiting him to a more observational role that only provided cursory glances into the youths' lives. Quite early on, Roks noticed how much time they spent on social networking sites such as Hyves (a formerly popular Dutch social media platform), Twitter, and Facebook, which they would browse via the center's computers or their smart phones. This peaked Roks's interest in their online worlds, and he began trying to locate their online profiles.
Roks's initial attempts at searching the youth's real names on social media were futile. However, a search of their street names and the names of their respective gangs led him to a few of their profiles, and a systematic search of their friends and follow lists steered him to the profiles of other community center youth. This proved to be a methodological breakthrough that granted him access to a lively digital world, as the youth shared copious amounts of information about their lives on their public social media profiles. Through studying their pages, Roks garnered significant insights about how the youth chose to portray their day-today whereabouts and happenings, school and leisure activities, and their interactions with others including police, which they represented through words, photos, and hashtags. The youth also frequently posted about their (alleged) criminal involvement and court cases. For example, Jack, a neighbourhood youth who always kept his distance from him, provided a detailed overview of his criminal case proceedings on Twitter, from the initial charge to the final court decision. Other youth posted images of themselves with what appeared to be various firearms and large sums of money in attempts to bolster their street credibility. In addition to using social media for reputation building, the youth also deployed social media for more instrumental activities such as the sale of drugs and stolen property or cybercrimes like phishing.
After his first strolls on the digital street (Lane 2018), Roks gradually decided to incorporate social media into his urban ethnographic research methodology. Roks's deployment of the One-Way Mirror Approach originally initiated as "cyber stealth" (Murthy 2008, 408), where he could "lurk" (Richman 2007) or "creep" (Trottier 2012) on the youth without their knowledge, which yielded notable insights into the performativity of street culture in a digital era. He did eventually send friend requests to his participants from his own social media profile, unmasking his online presence to the young men. Hence, although he initially adopted the One-Way Mirror Approach for his participants early on, as his project progressed, he eventually moved towards the Glass Window approach. In the same way that urban ethnographers often adapt their data collection efforts and presentations while in the physical field, new media scholars must acknowledge the amorphous nature of netnography and potentially adjust their approaches to data collection in the digital field as well. For example, changes to the popularity of different social media platforms (from Facebook to Twitter and eventually Instagram and Snapchat), and a growing awareness that police officers may be monitoring online accounts, resulted in many of Roks's participants utilizing private profiles.
This platform update meant that Roks could no longer secretly observe online dynamics but had to send friend and follow requests to his participants, which they would have to accept in order for him to be able to access their online worlds. Most of his participants accepted his requests and followed him back almost instantaneously. These common digital practices raise important questions about informed consent in fieldwork that navigates both online and offline spaces. By granting him access to their digital content, did his participants also provide their online consent for his study of their digital artefacts? Since he did not have to submit a formal proposal or get ethics approval 3 and existing scholarship did not engage with these questions, Roks had to rely upon his own ethical rules and considerations to establish guidelines for navigating the issue of informed consent on social media and disseminating associated data.
For example, during the first year of his research, Roks predominantly relied upon realworld ethnographic data in presentations and academic publications. In instances where he drew from his digital fieldwork, he only deployed online artefacts from participants who he also knew in the "real world," which ensured that "online-only" participants were not included. Additionally, like in the Glass Window approach, Roks often made his online presence known by liking his participants' posts and participating in comment sections. In turn, his participants engaged with his posts as well, sometimes even directly tagging his account. For example, one participant posted, "Soon, you can read about the all the things I did in a book, ask @Roks" on his account. As his research progressed, Roks used social media happenings to initiate conversations or group discussions about and on the physical street, for instance by asking clarification about specific terms in posts, in addition to using DMs (direct messages) to send interview requests. Though there is expansive literature on seeking informed consent for online research more broadly, urban ethnographers conducting fieldwork in the real and digital streets with the same participants may struggle to find relevance in these guidelines (c.f. Lane 2018). Hence, there remain notable questions on what informed consent entails for those conducting digital ethnography, which we address elsewhere (Urbanik et al. forthcoming).
As Roks's fieldwork progressed, he gained access to a different group of participants in the Forgotten Village-the Rollin 200 Crips. The Crips were generally older (30+ years old) and less internet savvy than the young men from the community center, with only a handful of them using social media. Consequently, his primary empirical data for this group came from traditional ethnographic activities-physically "being in the h200d"-as online research yielded little additional insights into their lives, as compared to the young men from the youth center. Thus, though accessing social media enabled data collection for the harder-to-access group of youth, and Roks had to navigate both the online and offline streets in his data collection efforts, this approach was far less relevant (and beneficial) for the Crips, despite the fact that both groups occupied the same physical streets.
Well before researchers initiate their fieldwork, the One-Way Mirror Approach allows them to monitor and analyze the minutiae of prospective participants' online presentations and interactions. For gang ethnographers, this may be particularly fruitful as street gangs have become increasingly visible and "searchable" online, and often proudly display indicators of gang loyalties, beefs, and even common hang-out locations. Perhaps equally important is social media's role in providing us with a basic understanding of who these individuals or groups are and how and what they choose to present, which may expedite our rapport building on both the physical and digital streets. Nevertheless, the benefits of the One-Way Mirror Approach may extend well beyond its usefulness in identifying desired participants during the initial stages of research, as some researchers may choose to utilize this method well after building rapport on the ground (see Van Hellemont 2012).
The Glass Window Approach: "Yo, You Got Instagram?" The other side of the field role continuum denotes research roles that are more participation based. 4 When applied to social media, this role can be described as the Glass Window Approach. The Glass Window Approach refers to researchers who share their social media accounts with participants, thus allowing for two-way visibility, scrutiny, and interaction. Urban ethnographers can learn about their participants online lives and participants can simultaneously learn about the researcher. This approach changes the participant-researcher relationship, in that instead of it being top-down like the One-Way Mirror Approach (we "see" our participants online, but our participants do not "see" us), this relationship becomes more lateral-we have access to what our participants post and they have access to what we share. 5 This approach also provides researchers with an additional "stage" for building and managing relationships and research identities with participants, 6 and affects the researcher's fieldwork in notable ways as it increases the researcher's "reach-ability" to participants (Pascoe 2009).
When interrogating how this approach relates to gang ethnography, it is imperative to analyze how our (prospective) participants' social status and power may influence their digital behaviours and consequently alter and/or impede our data collection efforts. Reich (2015, 400) notes "technology has made it much easier for potential subjects to scrutinize researchers." Undeniably, the Glass Window approach exposes the researcher to potentially greater scrutiny by (prospective) participants than the One-Way Mirror Approach or a limited digital footprint altogether. In Reich's study with high-status antivaccine parents, for example, (potential) participants researched her previous scholarship to uncover her politics and judge whether she could be trusted. In contrast, our participants-to the best of our knowledge-did not engage in a similar online "vetting" process apart from examining our social media platforms. This was likely tied to the fact that our participants had vetted our real or perceived identities, intentions, and politics in person, 7 which either afforded us access to them in the "real" world (Urbanik) or did not (Roks).
In contrast to Roks, Urbanik's use of social media in her research was premised upon the Glass Window approach. Despite never having considered that social media may play an important role in her ethnographic study at the onset of her research, Urbanik's field relationships led her to connect with her participants online rather early on, though her eventual fusion of urban and net ethnography was unplanned. Urbanik's participants had fleetingly mentioned their social media activities during her first summer of fieldwork (2013), but it wasn't until the following year that her participants increasingly began attuning her to the prominence of social media for their lived realities. One summer evening in 2014, Urbanik, Freestyle (17 years old) and Cuzzy (19 years old) were just hanging out outside Freestyle's house when, while playing on his phone, Freestyle turned to her asking, "Yo, you got Instagram?" When she responded that she did, he followed up with "Aight, well add me right now. Wait, I'll add you. What's your handle?" As Freestyle was adding her, Cuzzy pulled out his phone, searched for her account, clarified-"Is this you?-by showing her the profile, followed her, and insisted, "Yo, follow me back. Follow me back." This quick, casual, and seemingly trivial exchange digitally connected Urbanik and her core participants and granted her access to their online selfpresentations.
Initially, Urbanik was rather disinterested in her participants' online portrayals. The majority of their pictures, videos, comments, and tags could be characterized as performances of marginalized masculinity; the young men would post images of them showing off their new shoes, wearing brand name clothes, posing beside expensive cars, displaying wads of cash and gold chains/teeth, and making frequent mention to their "gangster-ness" and capacity for violence. Urbanik did not realize this important field development as such; at the time, young people often added each other on social media, and her relative age-proximity to her participants and unawareness of netnography made this seem uneventful. In fact, it was Freestyle and Cuzzy-both digital natives-who first attuned her to the value of their newly digitized relationship, as several days after this interaction, Cuzzy told her: "You better show all the whiteys [white people] at your school what real G's [gangsters] look like! Show 'em my page, so they know Toronto G's [gangsters] are real, not like the ones in Edmonton! [Urbanik's university city]". Cuzzy seemed eager that his social media page be deployed as evidence of his gangster-ness, and explicitly encouraged her to disseminate this new evidence at her university, something she had not previously considered and did not further interrogate until much later on in her research.
Similar to Roks's experiences, the two interactions described above spark additional debates about attaining consent online, specifically: participant-driven vs. researcher-sought consent. When digitally literate participants initiate online relationships with ethnographers that they know are studying their lives and to whom they have already provided an extraordinarily high level of consent to (by agreeing to be key participants), does this invitation not constitute participant-driven consent? By pointing out his account's potential relevance to Urbanik's study, did Cuzzy not provide informed consent? Further, we must consider whether such situations differ from common, everyday field processes. For example, participants often invite ethnographers into their homes or to attend parties with them. In these moments, we need not ask for consent to attend since we understand that the invitation itself constitutes consent. But is there something unique about the online world that mandates a separate, or additional layer of consent for key participants that goes above and beyond their broad-consent to serve as key participants?
Alternatively, how does or should the participant sample matter when determining whether consent was given to observe participants' online behaviours? May this be different for criminally-involved participants who generally tend to be more distrustful of anyone, including researchers, and who may therefore carefully vet who can see their online content? What about for criminally-involved participants who have public accounts despite their knowledge that anyone-including law enforcement-can easily find and access their digital content (Urbanik and Haggerty 2018;Urbanik et al. forthcoming)? Some gang ethnographers may be amused by the idea that despite having established enough rapport with participants to observe participants' real-world criminal activities and violence (i.e., drug and weapons trafficking, fights with rivals, etc.), they are expected to secure some sort of separate or additional consent to view their participants' online content upon being added by them. Even our participants may see some irony in these expectations. 8 Undeniably, things become more complicated when the ethnographer wishes to move the relationship online; consent is therefore initiated by the ethnographer and is "researcher-sought," which changes dynamics and raises different questions, as Roks described earlier (see also Lane 2018).
Despite her original disinterest, Urbanik slowly began to recognize that her online relationships with research participants were an invaluable research tool for understanding the social dynamics in the neighbourhood; the sociological and criminological relevance of their online presentations was immense as their online interactions notably shaped their real-world movements and behaviours (see Urbanik and Haggerty 2018;Urbanik forthcoming). Further, she could easily find and connect with many of her participants online through skimming existing followers and comments, she knew what was happening in the neighbourhood even when she was at home, and she was able to put faces to names for those she had not yet met in person. Urbanik's follows (for public accounts) and requests to follow (for private accounts) were usually met with her participants requesting to follow her back, and several participants requested to add her first. Urbanik used a higher level of caution for participants with private accounts who added her back, where she would have in-person conversations with them about whether this was a "personal" relationship or a "research" one. Though they thought the question itself was odd (likely because they actively approved or initiated the digital relationship), all participants expressed that the content could be used for Urbanik's research. However, social media did not emerge as a central feature of her analysis until her participants became increasingly concerned about how social media was inducing, aggravating, exacerbating, and accelerating neighbourhood violence, via gang and neighbourhood "beefs." As a result, Urbanik began paying greater attention to social media dynamics, connecting with more participants online, and also incorporated questions about social media into her interview schedule. It was only through relying upon participants' accounts and witnessing their presentations and dynamics online that she gained a more nuanced understanding of how social media was affecting street life, and was able to refine her research methodologies to capture different data. Equally importantly, the seemingly arbitrary decision to connect with her participants online, allowed her to maintain a constant "presence" in her participants' lives and keep up to date with neighbourhood happenings while away from the field. For example, though Urbanik was studying and working across the country from her fieldsite, social media kept her abreast of when her participants landed big rap gigs at prominent venues, released new music, or celebrated birthdays or Eid. For a more somber example, when someone is killed in Regent Park, Urbanik usually first finds out about the identity of the victim via "RIP" postings on social media, which work to inform others much sooner than in-person communication or news media channels. Moreover, recent updates to Instagram allow users to stream live videos, which allow her to track her participants' whereabouts, activities, and social interactions in real time. This change also required that Urbanik be versatile in how and when she checked social media, as (similar to Snapchat) content self-destructs after a specific amount of time, increasing the risk that she might miss important interactions. Hence, in the same way that flexibility is a key element of traditional urban ethnography, new media scholars need to be versatile in when, how, and why they use social media platforms in their research.
Further, although Urbanik was somewhat skeptical about sharing her own social media with participants given that she frequently posted about her own life happenings, several of her participants told her that if they ever suspected her of being an undercover police officer, her social media postings about her school, work, and social life, reinforced the fact that she was not an informant, but just "a regular girl"-like Whiz claimed. Several years later, she continues to follow and interact with her participants via social media, and her digital presence is so normalized that they frequently show her specific social media posts as they relate to neighbourhood and gang violence, oftentimes even decoding the posts for her to ensure she understands the post's "true" meaning. Urbanik still encounters iterations of the "Yo, you got Instagram? [or Facebook, or Snapchat]" question when she meets new research contacts, thereby continuously expanding her digital research network. However, while Urbanik is able to maintain some form of a fieldsite presence through social media, her physical absence undeniably alters the nature of her fieldwork and field relationships, as it limits her ability to compare and contextualize online dynamics with street happenings, and restricts her capacity to build/maintain rapport and otherwise manage field relationships to the digital realm.
Thus, the Glass Window Approach may provide researchers with additional ways to build (or damage) rapport with our research participants, as it allows us-if we desire-to keep in touch and track field developments 24/7, even if we are away. This approach also enables us to build (or lose) trust with our participants, as they can now more easily evaluate our lives and presentations, and better determine who we are outside of the field (i.e., not police). This approach to urban ethnography may also allow us to uphold our field relationships by sending direct messages, commenting on posts, but also, through more subtle means such as liking participants' posts. No longer are we limited to staying atop of developments via face-to-face interactions with participants or phone calls. Nevertheless, this approach also comes with significant risks as our online actions (i.e., friending or following a gang rival or personal contact in law enforcement) or inactions (i.e., not liking a participant's recent post/newest music video) can have just as tangible consequences for our field identities and relationships as our interactions in the "real world." Undeniably, the Glass Window Approach is much more difficult to navigate, though it may come with greater benefits (and risks) than the One-Way Mirror Approach.

"Keeping It Real" Online? Practical Issues
Both the field roles outlined above have immense potential in enriching urban ethnography's quest for capturing participants' lived realities. However, these approaches also produce novel practical considerations and ethical dilemmas. Based on our experiences using social media in ethnographic research, we outline three practical issues that are by no means exhaustive but provide a glossary impression of the questions and challenges of online activities during fieldwork.

How to Act and Identify as a Researcher Online?
First, we need to establish how to act and identify as a researcher on social media. Do we use fake accounts-arguably the "safest" bet for researchers? What about research-only accounts? If we use our "real" accounts, how much information about ourselves do we share with our participants online? Our participants can meet phantom or fake social media accounts with suspicion, a risk which may be particularly acute for scholars studying criminally-involved groups who are often weary of researchers being undercover police officers (Bourgois 2003). Fake accounts are usually detected by a lack of other friends, limited or non-individualized photographs, commentary, etc. In the same way many of us may be suspicious of accounts that appear surprisingly bare or otherwise look suspiciously impersonal, and may therefore decline friend/follow requests or limit permissions, research participants may also subscribe to similar safeguards to ensure that they are not sharing information with unknown/fraudulent/dangerous others (see also Lane 2018). Similar concerns may apply to research-only accounts. Though creating new or research-only social media profiles 9 may assist researchers in evading the ethical and personal dilemmas associated with sharing one's personal account, this approach might ultimately hinder our access and rapport-building efforts as participants may distrust our research identities and intentions. Hence, scholars must carefully balance the ethical and practical dilemmas associated with whether we should disclose our research identities and intentions online, and if so, how much will we share, or whether we might opt for covert research (or lurking)-like Roks, which can have important repercussions for the quality of our data, access, ethical dilemmas, and/or our safety (see Barratt 2012;Hine 2015;Kozinets 2010;Mkono 2012;Reid 1996;Sanders 2005, 71;Spender 1995). This speaks to an important yet challenging broader question: what does it mean for researchers to be real or fake with their participants?
Another potential limitation of sharing one's own personal social media account with participants is that it could further distance the actual or perceived social/economic/political distance between the researcher and her subjects. For example, if one's personal profile reflects one's privilege, immense cultural capital, vast racial differences in terms of family and friends, and one is studying impoverished, disenfranchised, and/or racialized groups, befriending participants on social media can further amplify their perceptions of a researcher's difference. Ethnographers have documented how their difference from participants actually enabled their access (Bucerius 2013;Urbanik 2018). However, these differences can potentially be mitigated through face-to-face interactions, whereas social media postings may serve as a continuous or uncontextualized reminder of positionality.
It is also important to note that the nature-and not just scope-of a researcher's profile may limit access to research participants. If a researcher's personal profile is littered with images, videos, or comments that differ quite vastly from those of the research subjects, this may potentially (though, not necessarily) harm the research relationship. This may be particularly true if a researcher presents themselves one way in the field and in a completely different way online. Inconsistences in self-presentations can spur suspicion about who we actually are. Since who we are (or how we present) as urban ethnographers is one of the primary reasons we can gain and maintain access to and build trust with our respective participants, we should not take these concerns lightly. Therefore, a researcher's increased visibility to research participants via social media must be managed carefully, or it runs the risk of being detrimental to our research relationships.
Another concern relates to the fact that our online presentations are complicated by the issue of "context collapse" (Marwick and boyd 2010). Prior to social media, our selfpresentations were often catered to distinct audiences, where we strategically presented different selves to different groups, based on what image we believed would place us in the best light for that audience (Goffman 1959). Yet the rise of social media has jeopardized our ability to control our self-presentations since they can now be broadcast to an infinite, yet uncontrolled audience (Yar 2005). Although this raises important concerns for many groups, including gang members and street involved populations, this can also present fundamental issues for urban ethnographers and gang scholars who actively engage with their participants online.
For example, Roks utilized the same social media account for his private life, professional life, and fieldwork, where participants were granted similar access to his information as family and acquaintances. While this uncalculated approach worked fine during data collection, the completion of the study years later and a falling-out with the leader of the Crips culminated in an in-person confrontation that ultimately pushed him to unfriend and even block certain participants. In hindsight, Roks would have opted for two distinctive accounts, separating his personal life from the online component of his fieldwork, keeping in mind the impression management considerations outlined earlier.
In contrast, Urbanik used two separate social media accounts; one for sharing her "real life" with family and close friends, and one exclusively "professional" account that she shared within academia. Her participants were all connected to her "real" account, and were thus privy to the same level of visibility that her close friends were. Given how Urbanik first digitally connected with Freestyle and Cuzzy, she did not make a strategic decision to add them to her private account-it just happened. However, she continued to use this account to connect with them and other participants later, as she recognized the added legitimacy that this established account (many friends, photos, dated yet also frequent interactions) afforded her during the early days of fieldwork when she was still working on trust-building. She could have granted her participants access to her professional profile only, but this likely would have made her appear less relatable, and the account's limited activity (few photos, far fewer friends, limited interactions) would not have eased her ability to build trust and establish herself as a researcher (and not a police officer/informant), but may have hindered her access. Because Urbanik had her participants on her private account, she was able strategically post images/ quotes/videos/memes that she strongly suspected would re-affirm her identity as a researcher and would make her appear more relatable to her participants. She would not have posted similar things on her professional account given the specific "front stage" professional self she was aiming to convey through that channel. This separation also allowed Urbanik to interact with her participants online freely, where she did not have to consider which pictures or videos she would like or comment on, out of concern that her academic colleagues may see and judge these interactions. This is particularly true for individuals who may not have understood that Urbanik sometimes adjusted her online activities to enable trust-building with her participants. Since presentations on social media are often strategic (Quinn and Papacharissi 2014), new media scholars should consider how their online presentations may affect their research access (see also, Reich 2015).
The Glass Window Approach also necessitates that scholars reflect on what type of "window" becomes available to their participants, and whether this "window" indeed results in greater transparency. Since Urbanik was a frequent social media poster, her participants had access to many snippets of her life, well beyond merely professional material and presentations. The Glass Window approach only affords two-way visibility if researchers have a notable social media presence, and when their social media presence paints a relatively holistic portrait of the researcher's life, moving beyond a strictly professional or purely researchstrategic presentation. The former option makes researchers more vulnerable than the latter; our participants can dissect our personal lives as played out online much to the same extent that we can analyze theirs. The latter does not allow for this greater two-way visibility; participants may "see" that a researcher attended a conference, but they may be limited from "seeing" their lives apart from professional milestones, which obfuscates transparency.

How Should we Make Sense of Online Data?
Notwithstanding the potential of incorporating social media into ethnographic research, we caution that urban ethnographers should exercise the same amount of thoroughness and triangulation when trying to unpack their participants' lived realities when examining their online presentations as they do in real life. Long before the omnipresence of social media, Erving Goffman noted (and even predicted) possible discrepancies between "an individual's virtual and actual identity" (1963,31). This is not just a play on words. Our participants-like us-may present themselves and their "realities" quite differently online, and so we must be extremely careful not to accept everything we see at face-value. This is especially true since what people say is not necessarily consistent with what they actually do (Jerolmack and Khan 2014), whether in the real or digital streets. Lane (2018, 170) highlights these concerns when he argues that since saying and doing are now situated both online and offline, urban ethnographers must acknowledge the consistencies and tensions between the two realms, which can only be done by examining both realms in tandem. If we accept either of these representations at face-value, we may miss important sociological insights.
For example, Urbanik encountered a situation where one neighbourhood "Old Head"-Chops (37 years old.)-vehemently argued that he would never be featured in neighbourhood rap videos circulated online because such affiliations could lead to his violent victimization (Urbanik and Haggerty 2018). Although Urbanik originally commended Chops for staying out of street politics, months later he was prominently featured in several new music videos circulating on social media. Had Urbanik not been actively following her participants online, she may have used Chops' purported abstinence from "reppin" in music videos as a prominent example of how neighbourhood men resist the temptations to support their local rappers by appearing in their promotional videos. Instead, by spotting Chops in several rap music videos, Urbanik could argue that even neighbourhood Old Heads find it difficult to resist the street credibility afforded to her participants via their participation in rap videos. Further, since Urbanik was able to compare what Chops told her with what she witnessed online, she was also able to question Chops about this contradiction, an exchange that uncovered additional insights about the pressures he experienced in supporting local rappers and gang members. Hence, "being online" infused Urbanik's ethnographic work on the ground with greater clarity than if she had restricted herself to traditional ethnography.
In an example from Roks's study, "being there" allowed him to better evaluate and contextualize what he witnessed online. Many of his participants would reference the Rollin 200 Crips in their online postings, often referring to the Forgotten Village as the "h200d." Among youth in The Hague, use of this term communicates territoriality, indicates particular insider knowledge about the Crips, and is used to demonstrate that the youth are streetwise and potentially, street-involved. Simon (16 years old), a youth who frequented the youth center, combined various symbols associated with the Crips in his self-presentation, both online and offline. He wore a blue cap, a blue bandana around his neck, an oversized blue t-shirt beneath a large blue hoody covered by an even bigger blue jacket, and had another blue bandana hanging out of the left back pocket of his sagging pants, matching the blue laces on his shoes. Simon's Twitter posts suggested that he spent significant amounts of time hanging out with the Crips in the h200d. For example, he often posted that he was "walking towards the h200d" or was "on h200d patrol," which combined with his physical appearance, suggested that Simon lived in the Forgotten Village and was a Crips member or affiliate. However, Roks's physical presence in the neighbourhood allowed him to conclude that neither were true; Simon did not live in the area, nor did he have any direct contact with members of the Rollin 200 Crips. Instead, through studying friends and social media accounts, Simon convincingly emulated the particular style, dress, and language of the Crips (Roks 2017). Thus, Roks's comparison of what he saw online, with what he witnessed on the streets, provided him with more comprehensive (and accurate) data.
As demonstrated, urban ethnographers who deploy netnography in their work should approach their findings from the street and the digital world with a healthy level of caution, paying particular attention to consistencies and inconsistences between these platforms. When examining gang-involved participants, we should be particularly careful in not assuming causal relationships between online performances and physical violence; though some interactions ultimately result in real-world violence, most do not (Stuart 2019). Since social media presentations are not necessarily "factual" representations of on the ground realities, and since on the ground representations are affected and complicated by the digital world, we need to adequately contextualize what we see online with what we see in the "real world," and viceversa. In order to do this well, urban ethnographers must deploy the same level of rigor, skepticism, and triangulation in our analysis about how these two worlds intersect, as they have traditionally done in standard neighbourhood-based ethnographies. Failing to do so will limit our ability to properly evaluate the data that we come across online.
However, making sense of online data can also be particularly challenging. This may be especially difficult for those of us studying criminally-involved groups and/or "gangster" culture, as their social media postings may be intentionally cryptic or misleading. Many street-involved groups are aware that the police monitor their accounts (Urbanik and Haggerty 2018), thus they go to great lengths to communicate secretly through the use of code words, slang, seemingly inconspicuous or senseless hashtags, the use of emojis, etc. Before we have a solid understanding of the field, our (prospective) participants, their street culture, and importantly-meanings of their online presentations and interactions-we may not be able to accurately determine what we are seeing, and what it suggests. For example, one of Urbanik's participants posted a story on Instagram that said "HMU 4 ." Without a solid understanding of the slang codes that her participants used online, she may not have understood that this post meant "Hit Me Up for Marijuana," and may have missed that this participant was trafficking drugs and advertising his product on social media.

How Should we Use Online Data?
The use of messages, images and videos on social media begs the question of how ethnographers can and should use this data. In particular, how can we protect our participants' (online) identities when using social media posts in our presentations and publications? Inputting Tweets, captions, and pictures into online search engines can sometimes directly lead us to the original poster's account, thereby compromising their anonymity and potentially subjecting them to harm. Should we further anonymize/alter Tweets or posts to avoid participants being tracked by law enforcement officials, rival (criminal) groups, or those who read our work? Undeniably, similar concerns about protecting confidentiality are a staple of "on the ground" fieldwork. However, these concerns pale in comparison to the ability for those without specialized knowledge or tools to trace posts to specific individuals within seconds.
For example, many of our participants shared images of weapons, bullets, cash, drugs, and stolen goods online. Though we have utilized such posts in our work or conference presentations, we both made several efforts to ensure that our use of these gang artefacts could not be traced to our participants' accounts, which may then betray their real names, their appearance, where they live, and who they hang out with. These techniques of digital anonymization included de-identifying Tweets, concealing all usernames (including those who may have liked or commented on posts), blurring display pictures, and otherwise removing/masking other features that could identify participants or the groups in question (i.e., hair styles, tattoos, license plates, gang signs/logos on attire). Urban ethnographers already aim to anonymize their participants as much as possible, so while concerns pertaining to protecting confidentiality are not necessarily novel, they do take on new forms when it comes to data mulled from the digital world.
Interestingly, as our respective participants became increasingly conscious of their online visibility and the risks associated therein, they began taking steps to conceal their identities from rivals, law enforcement, and unknown others. For example, Roks's participants began to blur their faces or would make their accounts private. Most of Urbanik's participants left their accounts open to the public, though they did begin to cover their faces with their palms, hats, ski masks, bandanas, or oversized hoodies, often claiming "no face, no case!" [If law enforcement cannot see your face, you cannot be charged/convicted]. These behaviors signaled a growing desire for anonymity, one that researchers must work to protect, and highlight that urban ethnographers should be cognizant that online presentations are fluid and can change quickly. Hence, urban ethnographers who utilize social media platforms should pay particular attention to changes in their participants' online presentations and interactions, as with much behavior in the "real world," such changes can reveal new sociological insights.
Further, urban ethnographers who wish to complement their analyses through including examples of participants' social media postings should be extremely cautious in ensuring that they do not jeopardize their participants' identities (Urbanik et al. forthcoming). In fact, the risks associated with using social media postings in publications are so severe, that in their study of Twitter postings from residents of high-violence Chicago neighborhoods, Blandfort et al. (2018) consulted domain experts to ensure that their annotation decisions, labels, and data dissemination were not searchable online and did not compromise posters' privacy. One way to ensure our participants' confidentiality is therefore to make any raw data we share "ungoogle-able," by altering the words, memes, contexts of images, photos, and/or emojis that our participants used, and also by not using any social media profile identifiers (Shklovski andVertesi 2012, see also Lane 2018: 184-186). We also need to be particularly careful that the totality of what we share in terms of data, publications, etc., ensures that while one piece secures anonymization, the same piece is not connected to other documents that betray it (Shklovski & Vertesi 2012). Hence, new media scholars should seriously weigh whether the inclusion of such material is truly necessary (Urbanik et al. forthcoming).
In this regard, three questions arise. The first question relates to ethical research practices and how scholars can secure Research Ethics Board (REB) approval for online data collection. Criminologists have previously outlined concerns and proposed suggestions relating to obtaining REB approval, securing informed consent, leveraging the data of non-participants, researcher self-disclosure, etc. (Haggerty forthcoming;Lane 2018). Elsewhere, we describe other ethical concerns relating to our respective projects and discuss REB issues, frustrations, and inconsistences (see Urbanik et al. forthcoming). What further complicates these matters is that "university ethics committees and IRBs [Institutional Review Boards] that govern research may not be well equipped in these areas, offering inconsistent advice across countries and institutions, not least because social media is an emerging, contentious, and somewhat unknown entity" (Urbanik et al. forthcoming). Such discrepancies in REB expectations across and even within the same institutions also unfortunately limit the usefulness/applicability of sharing best practices for dealing with REBs.
The second question speaks to debates about the value, necessity, ethics, and possibility-or lack thereof-of anonymization, both in the physical and digital streets (Allen 2015; Jerolmack and Murphy 2019; Shlovski and Vertesi 2013). Jerolmack and Murphy (2019) criticize the default option of masking our participants and field sites, which can reify ethnographic authority, falsely inflate the universality of the data, and hinder replicability. However, they also acknowledge that in certain cases, ethnographers have "an ethical duty to mask, even if not externally required by the IRB or their subjects, if they have a reasonable belief that revealing identifying information may present a tangible risk to subjects" (2017, 17). We posit gang ethnographies fit this bill, particularly given the legal risks associated with our participants' identities, neighbourhoods, and behaviors, concerns which are exacerbated by the blatant over-policing of disadvantaged and racialized men. For these studies, masking participants, if not field sites, is ethically and practically warranted or even mandatory, given our participants' particular vulnerability. These harms are especially fueled by the increasing role social media plays in policing, gang databases, and criminal trials (Behrman 2015; Lane 2018; Urbanik forthcoming), and by the fact that gang and criminally involved individuals occupy online and offline spaces that are excessively surveilled, in part relating to their socio-economic status and/or race (Patton et al. 2017).
The third question relates to whether it is ethical to modify online posts and gang artefacts, including our participants' statements and presentations. Though urban ethnographers commit to present data as accurately as possible, the surge in social media usage poses a significant risk to our participants' identifiability (Haggerty forthcoming). Hence, it is not only ethical, but we would argue, necessary to conceal these posts and identifying artifacts. In many ways, this approach is quite similar to what many urban ethnographers who anonymize their participants and field sites already do (i.e., altering demographic details, details of specific events, etc.). Nevertheless, it is imperative to note that as computer search technologies develop further and often to unpredictable degrees, we need to be forthcoming with participants about the fact that we may not be able to guarantee their anonymity (Jerolmack and Murphy 2019). The difficulty in protecting our participants' identities will only increase as media technologies and search engines continue to develop, so urban ethnographers deploying netnography in their research should consider consulting digital specialists to mitigate any potential risks.

Conclusion
The rise of the information age has sparked the "novelty" debate in the field of criminology, especially given the encroachment of criminal activities and behaviours into the digital realm. Grabosky (2001, 243) argues that "it has become trite to suggest that the convergence of computing and communications has begun to change the way we live, and the way we commit crime." However, he warns against the "overgeneralization and hyperbole that characterizes a great deal of discourse on the digital age." Although the online world, and social media in particular, have changed some facets of human interaction-and thus, also altered and presented new challenges/ opportunities for research methodology-we hope to have illustrated that we do not require entirely new methodological toolkits to grasp or examine our participants' online realities. In fact, our contribution shows that very similar field roles, dynamics, and considerations apply in the digital era, but that the novelty of some netnographic aspects warrants additional reflections, practical quandaries, and ethical questions.
We also want to emphasize that scholars can certainly conduct rich gang studies or urban ethnographies without incorporating social media into their work, so long as social media does not notably influence the phenomena under study. While being present in the streets and online allows us to more thoroughly analyze the relationship between these two domains and therefore produces superior understandings of our participants' lives, this methodological approach is only better for projects where the digital sphere informs offline interactions and processes, or where the digital sphere is of empirical interest itself. Hence, the importance and usefulness of social media in informing our ethnographic work is also dependent on several factors, including the research questions and respective samples. For example, a study examining the familial or romantic lives of gang members may not benefit from social media analysis (and one could conduct a rich ethnography using the good old shoe-leather approach), though a study examining neighbourhood "beefs" may miss key insights and evidence if limited to the physical streets. Alternatively, some gangs may have only a nominal online presence, if one at all, and social media may be less prevalent for incarcerated or ex-gang members. 10 Hence, we see social media as an indispensable tool of ethnographic fieldwork that is better suited for certain projects than for others.
Ideally, the research methods literature on fusing traditional urban ethnography with netnography-including this paper-would prescribe a normative stance that preferences specific approaches and practices over others and provides scholars with succinct guidance on how to navigate this new methodology. However, as we have demonstrated, there are numerous personal, contextual, temporal, platform-specific, and research-strategic variables that must be weighed in deciding whether and how a researcher should present themselves and behave online. Though we would generally advocate for privileging the more transparent Glass Window Approach over the One-Way Mirror Approach, the risks and benefits of this role are highly dependent on several factors, including: the participants themselves, the researcher's comfort level with sharing their personal/professional life with participants, their digital literacy, the content of their social media accounts and friends/followers lists, risks of betraying participant anonymity, age, gendered roles/expectations, and most certainly, their research identity, field relationships, and respective samples. Hence, though this approach may be beneficial for some scholars and some projects, it also presents significantly greater ethical and rapport-building risks than the One-Way Mirror Approach. The highly-subjective nature of this methodological approach-similar to the subjectivity that goes into traditional urban ethnographers' research identities and field decisions-therefore impedes suggestions for best practices. Nevertheless, we hope that the methodological and practical issues and concerns surrounding the incorporation of the digital street into traditional fieldwork that we have raised thus far ignite more rigorous scholarly debate on this complex and perhaps controversial, though undoubtedly valuable methodology.