Researching Gangs: How to Reach Hard-to-Reach Populations and Negotiate Tricky Issues in the Field

Abstract: Researching hard-to-reach populations which operate in spheres of illegality and violence is impacted by issues of rivalry, territorial inaccessibility, and distrust, as well as ethical and moral concerns. In this article I discuss the difficulties faced by female researchers carrying out fieldwork with gangs and focus on gender-based and race dynamics, which affect qualitative research in male-dominated spheres. Drawing on LEE's (1995) conceptualization of ambient and situational risks, I outline the risks related to conducting fieldwork with gangs in Trinidad and Tobago's violent outskirt areas. I furthermore stress the researcher's responsibility to survive fieldwork and draw on the privileges I enjoyed and the threats I faced as a young, white female. This study contributes to our understanding of how sampling techniques can successfully reach hard-to-reach populations in high-risk areas and within a limited time frame by introducing a refined sampling technique, the successive approach.

Key words: gangs; gender; safety; dangerous fieldwork; qualitative research; violence; conflict; Trinidad and Tobago; Caribbean; grounded theory methodology; expert interviews; comparative case studies

"When your friends die, you can never get accustomed to that. Your life is a mess when you involve in the life of crime. You see your friends dying. You see your enemies dying. You see that every day! The police want to kill you. Bad boys want to kill you. You killing people. So you know that eventually your time will come. So you just live fast and enjoy your life, and you don't make up your mind; by 25 or 30 latest, you dead" (Interview with an ex-gang member, Port of Spain, Trinidad).

What are gang members willing to fight, kill, and die for? Gang members become "soldiers" in what is often termed "gang war." Their commitment to killing and dying is the puzzle that drew me into researching gangs. Gangs are an intriguing, multifaceted character. They are brutally violent criminals, on the one hand, but perceive themselves as "Robin Hoods," defenders of communities, on the other. Qualitative research on gangs generates knowledge on the experience of gang members themselves (BICKFORD, 2008, p.209). Yet, the popularity of quantitative gang studies (BROTHERTON & BARRIOS, 2004, p.4; HUGHES, 2005, p.99) has ensured there is a dearth of empirical evidence about the gang phenomenon from the perspective of gang members. MOSER (2004, p.7) criticizes quantitative methodologies for failing to capture the phenomenon of urban violence and highlights the need for more research based on qualitative sociological and anthropological methodologies. By interviewing active gang members, gang associates, and residents in gang-controlled neighborhoods, I was able to gain a deeper understanding of why they felt that joining a gang was their only option left, how a career in crime became desirable, and how gang leaders become seen as legitimate community leaders who provide jobs, welfare, and protection. [1]

In this article, I elaborate on the dangers of undertaking such fieldwork, focusing on gender-based dilemmas female researchers face when conducting qualitative research in male-dominated spheres. The scarcity of qualitative research on gang members is related to issues of security and accessibility to the populations of interest. Gang members operate underground and often engage in violence, illegal trade, and other forms of crime. Therefore, scholars may face significant security risks as well as distrust and prejudices when trying to study these groups. Drawing on Raymond LEE's (1995) conceptualization of ambient and situational risks, I outline the risks related to conducting research with gangs, focusing especially on gender-related issues. I also elaborate on the researcher's responsibility to survive fieldwork and the advantages and disadvantages that female researchers have, sharing insights I gained from my fieldwork with gang members in violent no-go zones. The sampling technique presented here was tested during fieldwork in Trinidad and Tobago, which has the ninth-highest homicide rate worldwide, and contributes to our understanding of how sampling techniques can successfully reach hard-to-reach populations in high-risk areas and within a limited time frame. The fieldwork was based on a comparative case study (SEAWRIGHT & GERRING, 2008). Case studies are a powerful method to identify new hypothesis especially when causal complexity is present (GEORGE & BENNETT, 2005, p.19). I opted for an explorative study and a grounded theory approach instead of a theory-testing study because it bares the power to detect new thoughts, probable causes and underlying mechanisms that have been overlooked so far (CORBIN & STRAUSS, 1990; GLASER & STRAUSS, 1967). [2]

"How did you do that?" is usually the first question I hear about my research. The aim of sharing my experiences is to encourage young researchers to re-evaluate the (perceived) limitations of conducting field research themselves in order to gain new and relevant insights into social phenomena while ensuring their own personal safety. Conducting qualitative research with hard-to-reach populations can, depending on the type of population, pose an immense security risk. The dangerous nature of the task is linked to either the high-risk settings or the actors themselves (e.g., criminals, combatants, rebels, terrorists, or members of other violent collectives). Anthropologists, ethnographers, sociologists, and political scientists have all experienced the challenges of conducting qualitative research and the related hazards. The resulting publications have contributed to a growing body of literature, which covers "dangerous fieldwork" (LEE, 1995; NILAN, 2002), "danger in the field" (LEE-TREWEEK & LINKOGLE, 2000a), and "physical dangers to fieldworkers" (BELOUSOV et al., 2007). Research on communities in high-crime areas or war zones entails the threat of physical danger to qualitative researchers; however, it can provide valuable data on the social lives of people in communities which have few points of contact to the outer world. Researchers undertaking such work face physical, emotional, ethical, and professional dangers (LEE-TREWEEK & LINKOGLE, 2000a). In discussing physical dangers, LEE (1995, p.1) differentiates between "ambient risks" and "situational risks." Ambient risks stem from the dangerous nature of the research setting, where the researcher is "exposed to otherwise avoidable dangers simply from having to be in a dangerous setting for the research to be carried out" (p.3). Situational risks are linked to the researcher's presence or actions in the field, which may "evoke aggression, hostility, or violence from those within the setting" (pp.3-4). There is insufficient literature focusing on the dangers of field research for women in general and on ways to systematically minimize those hazards. Some scholars, however, have sought to address this. For instance, Vanda FELBAB-BROWN (2014, p.1), who has conducted research on criminality and militancy around the world, provides clear suggestions on how to minimize risks and underlines the fact that the "chance of surviving fieldwork in highly dangerous areas and on highly dangerous subjects" can be influenced by the researchers themselves. The responsibility to "survive" fieldwork lies in the hands of the researcher. Janet JAMIESON (2000, p.61) supports this claim, contending that researchers have been "reliant on their own experience, judgement and common sense" when facing dangers during fieldwork. Geraldine LEE-TREWEEK and Stephanie LINKOGLE (2000b, p.10) point out that qualitative researchers strive for excellence in the competitive academic world and are under pressure to "strike out" and "break new ground," but the "risks they take are frequently seen as exclusively their own." [3]

The process of researching gangs is affected by gang members' sensitivity to outsiders and by race and gender dynamics. The issue of gender in field research has typically been analyzed in terms of the advantages and disadvantages connected to it. Prevailing gender stereotypes and racial prejudices can impact a researcher's attempts to access and conduct research in the field. [4]

Problems arising during anthropological and sociological fieldwork have been examined in detail by several researchers (LEE-TREWEEK & LINKOGLE, 2000a; WARREN, 1988; WESTERMARLAND, 2000). It has been found that female researchers commonly face cultural expectations based on the assumptions that "women are weak, naive, unresourceful, and, in some settings, sexually provocative" (LEE, 1995, p.57). Agatha LISIAK (2015) has thoughtfully reflected on the appearance of the (female) researcher in the field and how the role of the researcher is expressed and influenced by the choice of dress (clothing) along with gender, age, class, sexuality, nationality and ethnicity. Catherine PALMER and Kirrilly THOMPSON (2010, p.421) draw on their fieldwork experiences with alcohol-based and male-dominated subcultures, and claim that researching spheres of patriarchal structures has particular impacts on how to proceed. Safety procedures included telephoning her administrative assistant whenever she arrived at an interviewed partner's place; if she didn't make this safety call within ten minutes after arrival, her assistant would call her, then the interview partner, and in case both of them didn't pick up, inform the police. PALMER and THOMPSON thus highlight powerfully the danger of physical harm. Heidi KASPAR and Sara LANDOLT (2016) discuss the topic of sexuality in fieldwork beyond the positioning of females as victims. They point out that sexuality and flirting can be strategically used by female researchers for their own benefits—thus turning the "disadvantage" of gender restraints into its polar opposite. Prevailing gender norms influence assumptions about the researcher and, in turn, how the research is conducted. Racial dynamics influence how a researcher is perceived and confronted. John GABRIEL (2000) or Caroline FARIA and Sharlene MOLLETT (2016) explain that "whiteness" can be a structural advantage which needs to be elaborated on. Meanwhile, feminist geographers have argued the need for a better understanding of the "powerful work of whiteness," which has until now gained little prominence in the political science literature on fieldwork methodologies (p.84). These considerations refer to privileges based on race, particularly that of being white. The argument is that white researchers need to elaborate on the privileges they enjoy thanks to "white privilege," which is not a personal but rather institutional set of benefits (KENDALL 2013 [2006], p.62). Thus, depending on the predominant gender norms and assumptions, stereotypes, and racial dynamics, being of the same sex, race, age, and national origin as those one seeks to interview can be advantageous or disadvantageous. [5]

Female-conducted qualitative research on gangs is scarce; thus gang research remains a male-dominated endeavor. Early qualitative gang research began with the Chicago School and Frederic THRASHER, and continued with well-known scholars, such as John HAGEDORN, Martin SÁNCHEZ-JANKOWSKI, David BROTHERTON and Luis BARRIOS, Scott H. DECKER, Malcolm KLEIN, Edward MAGUIRE, Dennis RODGERS, Henri MYRTTINEN, and Philippe BOURGEOIS, to name a few. Although female scholars are yet to gain prominence in the field of qualitative research on gangs, there are some exceptions, f.e., Sonja WOLF (2012, gangs in Central America) and Jennifer M. HAZEN (2010, gangs and rebel groups). Due to the male dominance of the field of gang research, literature featuring the experiences of female researchers is scarce. Even Lorine A. HUGHES's (2005) paper "Studying Youth Gangs" fails to address this issue, merely touching on the issue of gender with regard to female gang members. [6]

When interviewing "active" criminals, in contrast to former combatants or ex-members, establishing a rapport and trust with them is paramount; otherwise, the researcher's well-being is in jeopardy. Nissim COHEN and Tamar ARIELI (2011, p.425) identified these particularities in conflict environments, which are particularly difficult to conduct research due to misunderstandings, exaggerations, and an atmosphere of distrust. The environments where gangs are active are not typical conflict environments, but they offer valuable similarities to COHEN and ARIELI's particularities in conflict environments with regard to methodological considerations. For instance, COHEN and ARIELI explain that high levels of societal suspicion of researchers as outsiders directly impact the choice of research method employed. As in any fieldwork situation, challenges can emerge, such as cultural differences and language barriers. Yet conflict-affected situations also can also encompass political and ideological prejudices towards the researcher (e.g., anti-Western sentiments), technical accessibility (e.g., no-go zones), and fear and distrust (p.426). Gang rivalry contributes to distrust towards outsiders. Although neutrality is an option for a researcher, LEE (1995, p.23) states that maintaining neutrality in highly conflictual social relations is probably impossible. Further methodological issues that impact sampling and interview techniques are, for instance, a lack of accessibility, a lack of openness due to mistrust, security issues, non-representativeness, and bias (COHEN & ARIELI, 2011, p.426). Therefore, identifying, locating, contacting, and recruiting interview partners, as well as conducting the interviews, needs to be planned. [7]

Carrying out fieldwork in dangerous settings has various implications, ranging from sampling and accessing the population of interest, shaping the research agenda, and formulating strategies on how to manage potential risks. In this article I draw on the hazards of fieldwork, particularly of researching gangs, and the privileges and disadvantages of being a white female. I furthermore introduce a refined sampling technique, the successive approach, which is suitable for researching gangs, and I elaborate on particularities of qualitative gang research, such as ethical and moral considerations and race- and gender-based ambient and situational risks. [8]

My political science PhD thesis on the transformation of violence-prone groups involved three months of fieldwork in Timor-Leste in Southeast Asia and three months in Trinidad and Tobago in the Caribbean. I selected Timor-Leste and Trinidad and Tobago as two most different cases with regard to theories and assumptions about the phenomenon of (youth) violence (youth bulge, youth unemployment, experience of violence [war and conflict], high urbanization rates, poverty) (see BLATTMAN, 2010; FULLER, 1995; IDOKO, 2008; KURTENBACH, 2012; NORDÅS & DAVENPORT, 2013; PRZEWORSKI & TEUNE, 1970; URDAL, 2012). Most different cases are exploratory or confirmatory in nature and can eliminate rival explanations (SEAWRIGHT & GERRING, 2008, p.298). Case studies are very useful to explain how social phenomena come about. For instance, GEDDES (2010, p.129) asserts that case studies can identify variables, shed light on anomalies that existing theories cannot explain, and to this end contribute to the creation and revision of theories. Case studies are a powerful method to identify new hypotheses, especially when causal complexity is present (GEORGE & BENNETT, 2005, p.19). This article is based on a comparative case study (SEAWRIGHT & GERRING, 2008). In cross-cultural studies comparative research methods are useful to "identify, analyse and explain similarities and differences across societies" and, along this vein, to detect knowledge gaps and enhance new perspectives (HANTRAIS, 1995, n.p.). In this study I use a grounded theory approach that aims at describing and explaining relevant conditions and social processes (CORBIN & STRAUSS, 1990, p.5; GLASER & STRAUSS, 1967). In order to answer my research questions, I opt for an explorative study instead of a theory-testing study as I want to understand the phenomenon of gangs holistically. [9]

Both fieldwork endeavors were challenging, yet in different ways. In Timor-Leste, a country I had visited and worked in several times, the challenges I faced were mainly related to language (learning the local language Tetum to conduct interviews); transport and road safety (traveling on trucks and motorbikes), and health issues related to poor living conditions. Fieldwork challenges in Trinidad and Tobago, a country I had never visited prior to my study, were related to my personal safety in a highly violent and male-dominated sphere, which is the focus of this article. It is beyond the scope of this article to elaborate on both fieldwork experiences. Thus, because the aim is to elaborate on negotiating tricky issues in the field with a focus on security and access to hard-to-reach populations, I only focus on the fieldwork I conducted in Trinidad and Tobago. [10]

Trinidad and Tobago, a two-island state in the Caribbean and a well-known tourist-dependent country, has faced a major upsurge in violence, crime, and gang activity. During my time there, people in the streets randomly walked up to me to warn me about the dangers of the city, ranging from kidnapping to gang rape. These worrisome warnings were not unfounded: Trinidad and Tobago is afflicted by high levels of gang violence and women are particularly vulnerable. In this small country of 1.3 million inhabitants there were 379 murders in 2012, giving the country a homicide rate of 28.3 per 100,000 for that year (UNODC, 2013). In 2010 more than 75 per cent of all murders in Trinidad and Tobago were related to firearms (SEEPERSAD & BISSESSAR, 2013, p.10). Ethnic backgrounds are diverse in Trinidad and Tobago: East Indians account for 35 percent of the total population, Africans for 34 percent, and "mixed" heritage people for 23 percent (CENTRAL STATISTICAL OFFICE, 2011). Less than 0.6 percent of the population is Caucasian. Ethnic origins are reflected in party politics and areas of residence: those who live in the northern part of the island of Trinidad, especially in the underprivileged neighborhoods, are predominantly black. About 83 percent of the gangs are African, and 13 percent East Indian (KATZ & CHOATE, 2006). [11]

A war between two major gangs has pushed the homicide rates up and spread fear. These gangs have carved out territories for themselves, with Laventille on the outskirts of the capital Port of Spain, along with Beetham, Sea Lots, and Morvant being the most prominent no-go zones in the country. Laventille accounted for 23.8 per cent of all murders in 2005, giving it a homicide rate of 249 per 100,000 persons (MAGUIRE, WILLIS, SNIPES & GANTLEY, 2008, p.60). Getting access to these areas was a major challenge. The gangs have created unofficial borders for the geographical zones they control and have restricted freedom of movement for both regular citizens and gang members. The gangs secure their borders through snipers with high-power assault rifles, who are located at designated observation points. Invisible to outsiders, gang territory begins close to the capital city's major shopping street and main bus station. In the case of researching gangs in Trinidad and Tobago, ambient risks were becoming a victim of street crime, getting hit by a stray bullet during a shootout between rival gangs, getting shot while patrolling with the police or while riding with gangsters in a car who become the target of a drive-by shooting. Situational risks, in my case in this project were sexism, sexual harassment, discrimination, kidnapping, rape, or murder. The situation in Trinidad and Tobago was challenging due to the "no-go" areas, where violence is commonplace, and because the target group was difficult to locate and access. My physical appearance as a young, blonde, white female meant that I stood out in these areas and was confronted with assumptions people had about me. [12]

A (now deceased) community activist in Trinidad and Tobago who worked with youth and gangs for decades described the security situation, which is a reflection of how the violent potential of these groups is perceived from the outside. In terms of conducting research, he warned me to stay alert, but not get frightened:

"You have to remember, even though you are working with them [gangs], that they are monsters! Not because they are talking to you and negotiate with you, don't think that you dealing with a rational human being. No! ... I have to remember, I can't be frightened. I pretend I am not frightened. You cannot get frightened. You get frightened, you are dead. You cannot be afraid of them" (Interview with a community activist, El Soccorro, Trinidad). [13]

I followed his advice and made sure I was always in alert mode—which meant that although I was not afraid, I was always aware of the potential dangers of a situation. Being in alert mode allowed me to remain calm and assess how a situation was developing without panicking or becoming fearful. I learned to consciously control my emotions and body language. I employed confident body language, such as taking deep breaths, walking slowly, standing upright, sitting confidently yet relaxed, establishing eye contact, and approaching people and introducing myself without shyness. [14]

Access to a research setting can never be taken for granted (LEE, 1995, p.17). Access to a population of interest that engages in crime and violence is even more challenging. In the following I discuss the new successive approach I developed and employed to get into contact with gangs in Trinidad and Tobago. Based on purposive and snowball sampling, it also incorporates aspects of interviews with experts with contextual or insight knowledge (MEUSER & NAGEL, 1991). [15]

Experts are persons who have particular knowledge about a certain topic of interest. According to Michael MEUSER and Ulrike NAGEL (1991) experts are, in contrast to elites, persons who are part of a certain field of action related to the research topic (see BÜHRMANN, 2004). In a broader sense, anyone can be an expert on a certain topic. Experts are persons who have access to, or knowledge of, topics or processes through personal observation or own part-taking. Therefore, every expert might have a different opinion and position due to varying perspectives and observations. Jochen GLÄSER and Grit LAUDEL (2010, p.12) define experts as "a source of special knowledge about the social circumstances under study" (translated from German). MEUSER and NAGEL (1991, p.447) distinguish between contextual knowledge [Kontextwissen] and insight knowledge [Betriebswissen]. Experts with contextual knowledge are people working on or about a certain topic of our interest and who have a deep knowledge and understanding of processes and backgrounds, offering a good start at the beginning at the exploratory phase. Experts with insight knowledge are part of the processes under study and report from their own life. In this study, persons with insight knowledge are members, leaders or former members of gangs, which are considered as hard-to-reach populations in this study. In general, hard-to-reach or hidden populations may be illegalized or stigmatized or criminalized persons or those, who fear to be when revealing their identity. Illegalized populations can be illegal immigrants or illegal working populations, stigmatized populations may refer to drug addicts, homeless, homosexuals or prostitutes. Criminalized populations can be, for example, drug traffickers, religious extremists or terrorists, or gang members. Populations are also hard-to-reach when, for instance, there is no defined sampling frame (e.g., homeless people), persons who prefer not to be part of the population of interest due to a stigmatization of it (e.g., prostitution), persons with blocked accessibility (e.g., high political or wealthy elites), and persons who are hard-to-reach because they prefer not to be reached due to their operation in the underground and illicit activities, such as gangs and other illegalized collectives (MARPSAT & RAZAFINDRATSIMA, 2010, p.4). [16]

Qualitative and quantitative research methodologies have different assumption and pursue different research questions and goals. In quantitative studies, statistical validity in terms of representativeness is important and achieved by, for instance, a randomized sampling technique. In contrast, most qualitative studies that aim to generate theory don't have to rely on randomized samples as the understanding and deep analysis of a problem and its underlying mechanisms are the focus of the study. Qualitative sampling doesn't typically aim to be representative, as the researchers don't assume a "normal distribution" of what they are interested in and emphasizes that in qualitative research, respondents are not "as good as the next" as certain persons can provide valuable key insights and understandings while others don't (ABRAMS, 2010, p.539). In order to assess who has key insights, a high level of prior knowledge about a certain phenomenon and its connected population is necessary. This prior knowledge comes along with a certain degree of intention and judgment, as commonly referred to purposive sampling. Purposive sampling refers to "strategies in which the researcher exercises his or her judgment about who will provide the best perspective on the phenomenon of interest" (p.538). With regard to hard-to-reach populations, the logic of the purposive sampling is suitable, yet lacks precise instructions how to reach and select those persons who may provide the best perspective on the phenomenon. [17]

Researching hard-to-reach populations raises the question of which sampling technique to apply. One technique to access hard-to-reach populations of the category of indirect sampling is the time-location sampling (MARPSAT & RAZAFINDRATSIMA, 2010, p.5). The main idea is to identify places that are visited by the population of interests, such as shelters, youth centers or free healthcare facilities. In theory this technique allows for framing the population roughly by mapping out all places and note visit frequencies, yet it is flawed with numerous influences that can't be controlled for (e.g., wrong mapping of places, rapid changes of visit frequencies and neglecting the part of the population that doesn't visit these places) (ibid.). In this project it was not feasible since persons were first needed to facilitate the access to places where the population of interest socialized. Another technique that can be used to research hard-to-reach populations is the respondent driven sampling, which uses four initially identified persons who recruit a number of additional persons to fill out questionnaires (HECKATHORN, 1997). With the use of a recruiter, a coupons system and monetary incentives a certain sample size can be reached. The respondent driven sampling is similar to the idea of snowball sampling, which is a sampling method also used in conflict environments. Conflict environments, in which individuals or groups' interest are contradicted by the interests of other individuals or groups, are particularly difficult to conduct research in due to high levels of misunderstandings, exaggerations and an atmosphere of distrust and suspicion (COHEN & ARIELI, 2011, p.425). Environments where gangs are active are not typical conflict environments, but offer valuable similarities with regard to methodological considerations. For instance, COHEN and ARIELI explain that high levels of societal suspicion and distrust towards researchers as outsiders; have direct impacts for the choice of methods. The methodological challenges include identifying the population of interest, mapping their subjective perceptions, and understanding their needs, interests and concerns. As in any fieldwork situation, challenges such as cultural differences and language barriers can emerge, yet conflict affected situations also can also encompass political and ideological prejudices towards the researcher (e.g., anti-Western sentiments), technical accessibility (e.g., no-go-zones) and fear and distrust. Further key methodological issues that impact sampling and interview techniques are denied access, lack of openness due to mistrust, security issues, non-representativeness and bias (p.426). [18]

The snowball sampling technique is a non-probability technique that is very useful to get in touch with hard-to-reach populations. This chain-referral sampling is regarded as the most effective method to access hidden and hard-to-reach populations when a representative sample can't be created (COHEN & ARIELI, 2011). The advantage of the snowball sampling method is gaining initial knowledge, mapping research populations, testing first assumptions, refining sensitizing concepts and has practical advantage of generating familiarity, trust, and to this end, rapport. I strongly agree with COHEN and ARIELI who argue that "the belief that the researcher is acting in good faith is fundamental to the establishment of a working relationship" (p.428). The snowball sampling facilitates new contacts that were introduced by a trusted person. The limitation of this sampling method is the lack of representativeness due to the bias based on the selection of the first contacts. The bias can be intensified if the first contact person, the "anchor," is poorly selected and impacts the selection chain of further network contacts. This could be countered by, as Leo A. GOODMAN (1961) suggested, drawing a random sample of a finite population. As explained above, there is a difficulty of a finite population with regard to hidden and hard-to-reach populations, which means that drawing a random sample is impossible. Further problems with regard to a bias derived from the use of the snowball sampling technique; refers to the selection of gatekeepers. If a gatekeeper is used, further biases can emerge due to the gatekeeper's own network and possible personal prejudices towards the population of interest (e.g., a gatekeeper might only contact persons in locations where he/she feels secure due to his personal, religious, or cultural background). [19]

My decision to develop a refined sampling technique, which I named the successive approach, was influenced by the logic of snowball sampling and purposive sampling, methodological insights provided by researchers who have studied hard-to-reach population (accessibility difficulties, indefinite population) (MARPSAT & RAZAFINDRATSIMA, 2010), and the particularities of conflict environments (atmosphere of distrust and suspicion) (COHEN & ARIELI, 2011). [20]

The first step of the successive approach consists of identifying, mapping, and contacting individuals who work or live with the target population of interest. This entails purposive sampling, as researchers use their judgment to identify the individuals they think will provide the best insights. These individuals are similar to experts with context knowledge and are here referred to as "periphery persons." When sampling periphery persons, it is preferable to generate a sample with diversity of the roles as the perceptions of the research phenomenon might differ depending on the person's role (e.g., whether they are a teacher, religious representative, or social worker). The core of the first step is to build a relationship with these different persons, e.g., teachers, religious representatives, social workers. The successive approach uses purposive sampling to avoid incorrectly anchoring due to a diversity of access points (periphery persons). A diversity of perceptions can contribute to a critical understanding of the research topic. I mapped out people who were working with or on gangs in Trinidad and contacted them via telephone or email. I gave them a summary of my research intention, explained who I was, and asked if they would like to meet to share their views. Participation was completely voluntary, and individuals were only asked to share information they felt confident sharing. This meant that periphery people were more than points of recruitment; they were also interview partners with valuable context knowledge. Interview material derived from periphery persons can thus be a substantial part of data collection. This first phase of the successive approach is characterized by researchers trying to build a relationship of trust with the periphery persons. This gives the periphery persons the opportunity to get to know the researchers and their intentions. In turn, researchers are able to gain insights into the work with hard-to-reach populations and gain the trust of the periphery persons. This initial phase is extremely important in enabling researchers to become familiar with the research environment and to adjust to cultural differences. I familiarized myself with the environment by drawing a map with the streets, hills, and back alleys and color-coding gang territories to avoid accidentally strolling into a gang's turf. As gang-territory borders are difficult to detect for outsiders, I had to gather this information on territory from interviews with periphery persons, literature, and YouTube videos made by the gangs themselves, in which they prominently featured the street names of their territory. The familiarization phase included spending a lot of time close to gang-controlled neighborhoods—namely, in the sketchy business district in East Port of Spain, where a single street separates the shopping area from gang territory. This part of the familiarization phase also included using the chance I got to get into the gang territories. For instance, I was invited to join social workers who conduct community meetings all over Laventille and Beetham. I also attended a bible study group twice a week and joined them on a prison visit. In addition, I was invited to meet and talk to community members, and I joined a group of young doctors who had established a temporary medical check-up center. My familiarization efforts also meant that community members, including gang members, had noticed my presence. In fact, some gang members later told me that they had already seen me on earlier occasions. [21]

The second step of the successive approach involved contacting members of the hard-to-reach population. This step is based on the consolidation of trust between researchers and periphery persons. Typically, researchers ask their contacts if it would be possible to talk to certain persons of the hard-to-reach population, in this case gang members. In some cases, periphery persons might be unable to establish contact as gang members regularly switch their phone numbers, which makes tracing down their new numbers difficult. But according to my own experiences, however, the success rate is very high. The snowball sampling technique is theoretically a linear expansion from one point to another, including an exponentially increasing population. The successive approach starts with a larger population of periphery persons but does not increase exponentially. Usually, one periphery person comes up with one person of the hard-to-reach population. In my case this meant that I never asked a gang member to introduce me to another gang member. The successive approach is useful for researchers whose field research is limited to a short period of time, because it is based on parallel processes—that is, a high number of periphery persons are contacted at the same time and asked to facilitate contact with the target population. In this two-step approach periphery persons are at the center of two relationships of trust: with members of the populations of interest and with researchers, who trust that the periphery persons will not put their lives in danger by setting up dubious meetings. The bond of trust established between a researcher and a periphery person can enhance the researcher's security, as the periphery person develops a kind of responsibility to ensure the researcher's safety. A relationship between the researcher and the periphery person based on trust can enhance security based on the assumption that the periphery person will not place the researcher in harm's way by making bad decisions (e.g., arranging dubious meetings at night). More specifically, I noticed that the periphery persons felt responsible for my well-being and advised me when a situation was safe ("don't worry"), when to pay attention ("be careful"), when not to go to a meeting or leave a scene ("it's time to go"). In this vein, the successive approach creates a network in which a researcher can work with a certain level of personal safety.

Figure 1: Successive approach in two steps: Periphery persons and the hard-to-reach population [22]

In the case of Trinidad and Tobago, potential situational risks included street crime, sexual harassment, sexual assaults, kidnapping, rape, and murder. Personal rules can be useful in mitigating most risks but should be decided on a case-by-case basis. During my fieldwork, I 1. made sure I was always home before dark, 2. never disclosed my home address in Port of Spain, 3. stayed with a host family who would eventually notice if I went missing, 4. was honest about the research objective at all times, 5. did not drive in cars with gang members (due to the risk of a being caught in a drive-by shooting), 6. never carried valuables with me, and 7. dressed appropriately and refrained from sending the "wrong" signals" (e.g., allowing a gang leader to assume I wanted to meet him for a romantic date). LISIAK (2015) has elaborated on the issue of the right choice of dress that influences how the researcher is perceived. To further complicate this situation, I was around the same age as most gang members. I was often asked whether I had a boyfriend and told that I was beautiful. Female researchers face the clear situational risk of their interest in a respondent being misinterpreted as sexual desire, which could result in conflict. In my experience, it was a tightrope act to "lime" (hang out) with them in a friendly manner without sending messages that could be misinterpreted and then lead to me upsetting a gang leader's ego. In contrast to my own recommendation, flirting in the field can have a positive impact on conducting research, as Heidi KASPAR and Sara LANDOLT (2016) experienced. They share how they "played with their sexuality" and "maintained and deployed the sexualization" where they felt it was "safe and beneficiary" to their goals during fieldwork in Switzerland. The two feminist researchers justified their use of flirting by arguing that the position of the researcher is far less powerful than usually described in the literature (p.116). I clearly see the theoretical benefits of mingling, flirting, blending in, and building relationships of trust by collectively consuming alcohol or drugs, but this has to be decided by the researcher case by case. PALMER and THOMPSON (2010, pp.427-428) explained how the consumption of alcohol was their crucial point of entry into the culture of Australian Rules football fans and "fundamental to the research process in terms of access." In my case I declined alcoholic drinks and the many marijuana joints that were passed on to me for two reasons. First, although I knew I would have been able to develop a stronger bond with them by partaking, I wanted to uphold certain boundaries and have them see me as a researcher. Second, keeping in mind statements of gang leaders like "I have never tried ‘white meat' yet," I needed all my senses to detect a change in the atmosphere or the rapport or any sudden possible dangers; thus, being intoxicated did not seem to be a good idea. Gaëlle Rivard PICHÉ, who conducted fieldwork in El Salvador and Haiti, similarly stated that she, as a woman, could not meet up with gang members for beers in the evening for safety reasons (TEMPERA, 2014). This points out to the power relations related to gender, sexuality, whiteness, ethnicity, and age surface as they tremendously influence the relationship between the researcher and the researched. [24]

I also had to reflect on the prevailing racial dynamics in Trinidad and Tobago. White Trinidadians descended from early settlers—enslavers and plantation owners—and today remain associated with the upper class. Based on my experience, they were perceived as conceited, racist, and elitist. These assumptions about white people influenced how my research participants thought about me. My ethnic background was commented on frequently, and in Laventille I was told that they hardly ever see white women in their area. At the same time, I noticed that they appreciated that I came and that they wanted to make me feel comfortable. On one occasion, I met a gang leader in his social housing apartment in Laventille. When we sat down on the stairs, with views over the neglected and impoverished sheds of Laventille, he looked at me. Before the interview started he suddenly pointed out that we were similar despite the racial differences. He said: "Although you are white and I am black, we have the same blood underneath our skin" (Interview with a gang leader, Laventille). [25]

My appearance as a young white woman in a male-dominated and violent sphere was predicted to be a disadvantageous combination, rendering me a "super-victim." But Patricia RAWLINSON (2008 [2000], p.299) asserted that "in disadvantage lies its polar opposite," as female researchers ("outsiders") can enjoy a certain level of respect. LEE (1995) described research settings in which male researchers were threatened or beaten up for their alleged membership in a rival group. Gangs in Trinidad are predominantly composed of African men. In my case as a white woman I was obviously not a gang member. I did not perceive my physical appearance as a disadvantage at all, as many people had predicted; in fact, it was clearly the opposite. My experience supports Carol WARREN and Paul RASMUSSEN's (1977, p.366) contention that in certain situations of male authority, females are regarded with less initial suspicion; and Candice ORTBALS and Meg RINCKER's (2009) argument that prevailing gender norms may aid females in their research access and success. I was a welcome break to their everyday routine, and by the way they talked to me, I knew that they did not perceive me as a threat ("Let me tell you something, sweetie!"). My physical appearance supported my standpoint of an outsider who was not involved in the ongoing conflict. [26]

Researching groups that are linked to violence and crime calls for special attention. The security and well-being of the researcher and the researched is of major importance. In this project I interviewed persons according to their membership in an illegalized group. In Trinidad and Tobago, as in many other countries, gangs are illegal and membership in these groups is prohibited by law and fined. According to Trinidad and Tobago's anti-gang law, "gangs are unlawful" and anyone who is "a member of a gang" or "telling anyone that he is a gang member or suggesting to anyone that he is a gang member commits an offence and is liable on summary conviction to imprisonment for ten years" (REPUBLIC OF TRINIDAD AND TOBAGO, 2011, p.5). Furthermore, "a person who is a gang leader commits an offence and is liable on conviction ... to imprisonment for twenty-five years" (ibid.). This legal context had various impacts on my strategy of how to conduct my research, especially with regard to maintaining interview partners' integrity, safety, and well-being. I had to ensure that there would be no negative repercussions (e.g., legal prosecution) for my interview partners. First, I let my interview partners decide on the level of anonymity used. Thus, while some openly agreed to be recorded for the interview and revealed their real names, some did not want to be recorded and only gave their street names. Second, I refrained from asking any questions (e.g., on gang affiliation, murder, other crimes, or weapons) that would put them in an uncomfortable position. One young "shotta" (a gang member who is notorious for many killings) admitted that he was nervous when we started talking. I suggested to him that he could just smile if I asked anything he did not want to talk about, and we would continue with another topic. This option was a big relief for him, knowing that there was a "backdoor" to avoid to be cornered. Some of the young men I talked to had no experience with interviews and were understandably reluctant to talk to me out of fear that I could be an informant working for intelligence agencies such as the Federal Bureau of Investigation (FBI). I provided them with my business card so that they could Google me and double-check that I was who I said I was. The provision of personal information became a bargaining chip with which to break the "unidirectional flux of information" (KASPAR & LANDOLT, 2016, p.108). Being honest is of major importance. A community activist explained that gang members should not be underestimated: "They will check you out. They test you. You have to be genuine. They are very, very bright. They evaluate you, and if they didn't like you, they wouldn't be with you" (Interview with a community activist, El Soccorro, Trinidad). [27]

Researchers require a certain level of social competence in order to sense which topics their interview partners want to avoid and which ones they love to share. For my interviews, I never showed up with my questionnaire; rather, I used a kick-off question ("How did Laventille become the way it is now?") which got the conversation started. My question worked very well because it was precise yet open. I realized that everybody knew what I meant when talking about Laventille, an area which had become synonymous with violence, crime, and gangs. I could talk about the topic of interest without having to directly speak about the "elephant in the room" or use offensive words such as "gang leader" (they regard themselves as "community leaders"). The kick-off question also allowed them to elaborate on whatever they thought was relevant. My interviews with gang members consisted of listening to what they had to share and then following up questions with questions that took us deeper into the conversation. [28]

The awareness of ambient risks contributes to developing a conscious strategy of how to minimize or avoid them. SAMPSON and THOMAS (2003) identified the sinking of the vessel they conducted their research on as an ambient risk. In PALMER and THOMPSON's (2010) case, ambient risks were the threat of robbery and violence at football grounds. Ambient risks of researching gangs relate to gang rivalries and entering territories divided according to invisible borders. Yet, there is ambivalence between real and perceived risks. The areas where gangs are active are usually considered no-go zones and hot spots of crime and violence. Unfortunately, heavy restrictions on freedom of movement in the deprived areas of Laventille, Beetham, Sea Lots, and Morvant made it very difficult to meet interview partners in these areas. They refused to leave their territories out of fear of being attacked by a rival gang once they left safe ground. This meant that it was not possible to conduct the interviews in neutral settings. In such instances, researchers must be willing to enter high-risk territories and face the challenges connected to doing so. I met with gang leaders in different places in their natural environment, where they usually "lime" (hang out)—for instance, in open staircases of social-housing buildings (referred to as plannings), on porches, or outside on walls or plastic chairs, at times with stunning views over the hills of Laventille and the ocean. [29]

A researcher must not be naïve when choosing a location, but distinguish between real and perceived threats and deal with the ambient risks. For instance, FELBAB-BROWN (2014, p.11) claims "the greatest dangers of sexual violence ... [were] not at the hands of a criminal or insurgent [she] interviewed but on an overnight public train to Calcutta." LEE (1995, p.15) found that persons in violent settings may even "welcome the researcher who can provide them with an audience and voice." This underlines the ambivalence of dealing with violent criminals whose willingness to use violence is not necessarily directed at the researcher. A gang member I interviewed complained about the media depicting his neighborhood as a "war zone" where people get killed:

"You come here, but the businesses would never come here. And when they come and see it, it's a different scene to how the media have it! The media have it like a war zone nah and people losing their lives, people just go and kill, but it's not so!" (Interview with a gang member, Laventille, Trinidad) [30]

From their point of view, the media "intentionally misguides" (Interview with gang leader, Port of Spain, Trinidad). At the time of research, there were two daily television shows that focused exclusively on crimes in Trinidad and Tobago: Crime Watch hosted by Ian ALLEYNE and Beyond the Tape hosted by Roger ALEXANDER, a police inspector. These shows cover all types of recent crime in the country, such as robberies, car accidents, domestic violence, rape, child molestation, murder, white collar crime, organized crime, and gang violence. They help to create the devilish public image of areas such as Laventille and contribute to the fear of the average citizen. Understanding the media culture of the country in which fieldwork takes place can contribute to a more differentiated picture. In my experience I realized that violence, murder, and other forms of crime were cultivated and featured prominently on Trinidadian media channels, which can be removed from reality. Similarly, a former gang member argued that the media focuses on the negative headlines, which only contributes to the fear of regular citizens. To this end, the media creates an image which is far different from reality, as even those who are involved in crime are not "crazy people" who randomly attack people (Interview with ex-gang member, Port of Spain, Trinidad). He explained that gang members are not mentally ill people who act on a random basis and that the real risks are limited to shootings:

"Plenty government agencies are frightened because they think that once you get into these communities they [the gangs] will attack you and rape you and shoot and rob you, but it's far different from that! ... They are not maniacs, they are not mad people, the most dangerous thing they have is when they have a shootout in the community and you are there, a stray bullet you might pick up" (Interview with an ex-gang member, Port of Spain, Trinidad). [31]

Similarly, another gang leader emphasized the peacefulness in his neighborhood and failed to understand why people from outside of his community are afraid to come and, if they do, why they only come with police escorts:

"Well you see now, you see like how you come here by yourself? Some of the people who come, come here with the police because they are frightened. I don't know what they are frightened for. ... If they come, they come with lots of police, for no reason! I ask them, why you come with all this police?" (Interview with a gang leader, Laventille, Trinidad) [32]

HUGHES (2005) picked up on this perspective in her paper "Studying Youth Gangs," pointing out that youth gangs don't shoot and kill all the time. Their everyday activities are "hanging out," "doing nothing," or "drinking and partying" (p.103). To this end, researchers must evaluate whether they face becoming a collateral damage victim or the direct target of violence. [33]

My analysis of the situation also showed that the communities had been stigmatized for decades. It was a surprise to me that many born-and-raised Trinidadians, including scholars and ministerial workers concerned with security issues, had never set a foot in the areas but bore prejudices against them. In the view of gang members and regular citizens from gang-controlled territories, the Trinidadian society excludes and stigmatizes people from Laventille and its surroundings, treating them as inferior and "paint[ing] them with a broad brush." People from these areas face difficulties in finding employment outside of their communities because they are perceived as "bad fellas," "gang members," or criminals from low-income areas who are "thieving, murdering, and coming to destabilize" (Interview with steel band manager, East Port of Spain, Trinidad).

"So the society has created their own problems with the people of East Port of Spain by victimizing them, treating them as second class citizens, making them feel unwanted, making them become inferior" (Interview with a resident on Duncan Street, East Port of Spain, Trinidad). [34]

In addition to being stigmatized, Laventille is politically, socially, and economically isolated and neglected. A pastor admitted that NGOs and churches have "sold out to the gangs because of fear" (Interview with a pastor, Maraval, Trinidad). Thus gangs have "filled the void" that has been left by society's leaders and politicians.

"So the void and the gap that has been left, somebody will fill it. And right now the person who is filling the void that the leaders of society and politicians have left in these areas, are the gangsters. The head of the gang is filling it by giving people money for food, clothes ... at the end of the day when people leave the church they go home hungry without money and the gangster might give them food and money! What do you want people to do? This is the reality!" (Interview with a resident on Duncan Street, East Port of Spain, Trinidad) [35]

By interviewing gang members, I was able to grasp how they perceived themselves: as victims of capitalism and state corruption and, consequently, as the Robin Hoods of their stigmatized communities. Their frustration with politicians and growing political apathy is based on the perception that the corrupt political elite lives at the expense of Trinidadian society. Political neglect was a common narrative as a major contributing factor to the emergence of crime.

"With the violence and the killings ... it have this kind of way... but it come down to one thing: the political figures in the country is not dealing with we [us]! And that's what's going on, sweetie" (Interview with a gang leader, Beetham, Trinidad). [36]

A leader told me that he grew up in poverty and that politicians steal from society to "get themselves rich" and that "the rich people need to share some of the money":

"This place needs fixing. Look at how nice London is! Look at how nice Germany is! So why we [our] place can't be nice too with all the money we have in this country?" (Interview with a gang leader, Laventille, Trinidad) [37]

While certain parts of the society in Trinidad and Tobago live in affluence, other parts are deprived of basic needs, such as sanitation infrastructure. During my stay in Trinidad, I noticed a geographical segregation between gated communities with expensive American-styled villas secured by armed private security companies, on the one hand, and the underprivileged neighborhoods of Laventille and the Beetham, the latter sandwiched between highways and a foul-smelling landfill and characterized by wooden sheds and broken houses, on the other.

"We live here in Laventille and we still have people living without toilets, [they only have] shit holes, latrines ... plenty latrines around here. And the government hold on to the money. You know they thief [steal]! Everybody thieving this country. I really don't know what they are doing with the money; I can't say" (Interview with a gang leader, Laventille, Trinidad). [38]

A young "shotta" explained that "the government like [sic] the crime." Another leader similarly argued that certain parts of Trinidadian society are purposefully kept in a state of dependency so that they can be controlled.

"Trinidad have [sic] enough work for the whole population that is unemployed. It have [sic] enough work. It have [sic] enough money to run this country and make everybody feel like somebody to a point. I don't mean like a lavish lifestyle, to buy a boat, car, plane, millions of dollars. I mean dealing with the family and that" (Interview with a gang leader, Beetham, Trinidad). [39]

In my interviews I learned that gangs drum up business and jobs for the community through social work programs and government contracts. The role of the gang leader is important as the creation of employment depends on it. Through companies and NGOs, the groups manage to secure funds (such as government contracts) for infrastructure projects that they can coordinate and which can provide work for their men. Gangs thus represent the largest "employer" in most stigmatized areas. The construction sites, such as houses, stadiums, roads and drains, were frequently pointed out to me, which highlights an important aspect of fieldwork: I was able to triangulate statements made by periphery persons/experts with context knowledge and actively involved persons (experts with insight knowledge) with my own observations. Conducting interviews in these areas also provided me with the opportunity to observe the social interactions between gangs and community members. I found that gang leaders provide social welfare, food, and financial support to single mothers and poor people and also send children to school. In their own narrative they were legitimate community leaders. One social worker described them as informal support systems. They were also seen as father figures.

"Many of the gang leader are virtual fathers. And that is what has confirmed and reinforced them in the society. It is serious! (...) they are the mentors!" (Interview with a pastor, Maraval, Trinidad) [40]

Interview partners stated that gangs provide what would be considered social welfare to the poor and financially deprived people. A statement made by a gang leader supports this claim.

"If I see people hungry, I feed them. I say, in my community like all over the Trinidad, we have the poor. Help people, that my joy. Yeah" (Interview with a gang leader, Laventille, Trinidad). [41]

I had met with this particular gang leader on several occasions. He sometimes called me on my cell phone to chat. He was kind and polite. When I met him in Laventille, I was able to observe him in his community. When I heard that many teachers are unwilling to teach in these underprivileged areas, I asked him if there were any classes at the community center in his neighborhood. He said:

"Not right now. I am trying to get some classes and thing to go on in there. I want a teacher for the slow learners. A teacher with lots of patience for the slow learners. And a cooking class" (Interview with a gang leader, Laventille, Trinidad). [42]

My own observations support these statements. For instance, I spoke with a former prisoner who used crutches who explained how the gang leader provided him with money and clothes after his release. During my meet-ups with the church group, I observed the gang members play a ball game with the local kids and reminded them to be on time for the church meetings. During interviews, I was also told that gangs provided protection and set rules and regulations. I witnessed this during an interview with a gang leader, who was approached by a crying girl whose necklace had just been robbed. She asked the gang leader for assistance in finding the thieves. The gang leader sent out some of his men to find the thieves and explained to me that the chances of finding them were high, while "the police has no chance." The two thieves were eventually identified. I asked the leader what he would do with them. "Sometimes I beat them up," he said, "but not this time. But rapists get killed" (Interview with a gang leader, Laventille, Trinidad). [43]

In this article I discussed the difficulties faced by researchers carrying out fieldwork with gangs. Qualitative studies based on interviews with gang members are limited, yet these populations can be a tremendous source of information and provide rare insights. [44]

Researching hard-to-reach populations calls for particular considerations, especially when the research is to be conducted in spheres of illegality and crime. Ethical considerations also have an impact on the research strategy, especially with regard to maintaining interview partners' integrity, safety, and well-being. Researching active gang members can put the researcher in the uncomfortable position of witnessing, violence and crime and illegality which can turn the researcher's role from that of "outsider" to that of unwilling witness. Drawing on LEE's (1995) conceptualization of ambient and situational risks, this article outlined the risks related to conducting fieldwork with gangs in Trinidad and Tobago, where gangs are active within territories usually considered no-go zones or hot spots of crime and violence. The ambient risks of this research were linked to territory and rivalry and consisted of becoming a victim of street crime or getting shot during a shootout between rival gangs, while patrolling with the police, or while riding in a car with gang members, who become the target of a drive-by shooting. Situational risks in this project were sexism, sexual harassment, kidnapping, rape, and murder. The successive approach technique I employed, is based on the logic of snowball sampling and purposive sampling, entails developing relationships peripheral persons, who can facilitate safe access for the researcher. [45]

I was able to develop a multifaceted perspective on the gang phenomenon and learned that gangs are indispensable social and political actors that are rooted in between the government and civil society. They have created a parallel structure with their own rules and regulations, an underground economy, and legal businesses and perceive themselves as victims of injustice, which legitimizes their brutality and role as defenders of communities. The line between legitimate community leaders—who are recognized as such by politicians and the police—and criminal actors is blurry. My immersion in Trinidad's gang culture allowed me to see what roles gang leaders play in their communities, to learn how gangs have developed over generations in the areas of Laventille, and to develop a better understanding of what hundreds of young men are willing to fight and die for. [46]

Against the backdrop of the dangers associated with carrying out fieldwork, I call for researchers to thoroughly evaluate their situations, as real and perceived security threats may vary. I focused on the particularities of conducting qualitative research on gangs and gender-based dilemmas for female researchers undertaking such work in male-dominated spheres. There is a need to further reflect on the relationships between researchers and researched populations with regard to race and gender and outline the privileges and disadvantages associated with each. The intersection of race and gender can position a female researcher as a potential sexual object (beyond her choice) or, depending on the research context, can be a key to success by securing her access and trust. Some may question whether conducting fieldwork in such high-risk environments is worth the risk. I argue that in a competitive publishing environment, researchers must consider the costs and benefits of carrying out dangerous fieldwork. Such work can provide the researcher with the chance to gather new insights into the life of one of the world's under-researched populations by providing such groups with opportunities to share their points of view, thus resulting in a multifaceted picture. However, there is the danger that some researchers may feel obliged to conduct as many interviews as possible to remain competitive in their academic field. The pressure is increased when the field research has already been financially funded. This pressure to deliver may push some researchers beyond their boundaries of personal safety. With this in mind, senior researchers, supervisors, institute directors, and university professors alike should take a stronger stance against the "numbers game." Exiting the field should not be regarded as a failure but as a viable option. Supervisors should better support their doctoral students in preparing (e.g., through security training and workshops on field research) and conducting research (e.g., through support networks and emergency contact persons). These realities of research have been largely overlooked. [47]

This article was written during my doctoral studies at the GIGA German Institute of Global and Area Studies, Hamburg. The field trips were financed by the German Academic Exchange Service (DAAD). I am grateful to Jessica WILCZAK for her valuable comments on an earlier version of this article.

Abrams, Laura S. (2010). Sampling "hard to reach" populations in qualitative research: The case of incarcerated youth. Qualitative Social Work, 9(4), 536-550.

Belousov, Konstantin; Horlick-Jones, Tom; Bloor, Michael; Gilinskiy, Yakov; Golbert, Valtenin; Kostikovsky, Yakov & Pentsov, Dmitri (2007). Any port in a storm: Fieldwork difficulties in dangerous and crisis-ridden settings. Qualitative Research, 7(2), 155-175.

Bickford, Annette (2008). Research methods. In Louis Kontos & David Brotherton (Eds.), Encyclopedia of gangs (pp.208-210). Westport, CT: Greenwood Press.

Brotherton, David & Barrios, Luis (2004). The almighty Latin king and queen nation street politics and the transformation of a New York city gang. New York: Columbia University Press.

Bührmann, Andrea D. (2004). Review: Jochen Gläser & Grit Laudel (2004). Experteninterviews und qualitative Inhaltsanalyse [The expert interview and content analysis]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(2). Art. 21, http://dx.doi.org/10.17169/fqs-6.2.476 [Accessed: October 10, 2017].

Cohen, Nissim & Arieli, Tamar (2011). Field research in conflict environments: Methodological challenges and snowball sampling. Journal of Peace Research, 48(4), 423-435.

Corbin, Juliet M. & Strauss, Anselm L. (1990). Grounded theory research: Procedures, canons, and evaluative criteria. Qualitative Sociology, 13(1), 3-21.

Faria, Caroline & Mollett, Sharlene (2016). Critical feminist reflexivity and the politics of whiteness in the "field". Gender, Place & Culture, 23(1), 79-93.

Fuller, Gary (1995). The demographic backdrop to ethnic conflict. A geographic overview. In Central Intelligence Agency (Ed.), The challenge of ethnic conflict to national and international order in the 1990’s (pp151-154). Washington D.C.: CIA.

Gabriel, John (2000). Whiteness. Endangered knowledges, endangered species?. In Geraldine Lee-Treweek & Stephanie Linkogle (Eds.), Danger in the field: risk and ethics in social research (pp.168-180). London: Routledge.

Gläser, Jochen & Laudel, Grit (2010). Experteninterviews und qualitative Inhaltsanalyse: Als Instrumente rekonstruierender Untersuchungen. Wiesbaden: VS Verlag für Sozialwissenschaften.

Glaser, Barney G. & Strauss, Anselm L. (1967). The discovery of grounded theory: Strategies for qualitative research. Chicago: Aldine Pub. Co.

Geddes, Barbara (2010). Paradigms and sand castles: Theory building and research design in comparative politics. Ann Arbor, MI: University of Michigan Press.

George, Alexander L. & Bennett, Andrew (2005). Case studies and theory development in the social sciences. Cambridge, MA: MIT Press.

Goodman, Leo A. (1961). Snowball sampling. The Annals of Mathematical Statistics, 32(1), 148-170.

Hazen, Jennifer M. (2010). Understanding gangs as armed groups. International Review of the Red Cross, 92(878), 369-386.

Heckathorn, Douglas D. (1997). Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems, 44(2), 174-199.

Hughes, Lorine A. (2005). Studying youth gangs: Alternative methods and conclusions. Journal of Contemporary Criminal Justice, 21(2), 98-119.

Idoko, Nick (2008). Nigeria: Youth violence and the challenges in the age of globalization. In Corinna Hauswedell & Sabine Kurtenbach (Eds.), In war as in peace: Youth violence, a challenge for international co-operation = Im Krieg wie im Frieden? Jugendgewalt als Herausforderung für internationale Kooperation (pp.137-156). Rehburg-Loccum: Evangelische Akademie Loccum.

Jamieson, Janet (2000). Negotiating danger in fieldwork on crime. In Geraldine Lee-Treweek & Stephanie Linkogle (Eds.), Danger in the field: Risk and ethics in social research (pp.61-71). London: Routledge.

Kaspar, Heidi & Landolt, Sara (2016). Flirting in the field: Shifting positionalities and power relations in innocuous sexualisations of research encounters. Gender, Place & Culture, 23(1), 107-119.

Kendall, Frances (2013[2006]). Understanding white privilege: Creating pathways to authentic relationships across race. New York: Routledge.

Kurtenbach, Sabine (2012). Jugendproteste – blockierte Statuspassagen als einigendes Band. Aus Politik und Zeitgeschichte, 62(25-26), 48-54.

Lee-Treweek, Geraldine & Linkogle, Stephanie (Eds.) (2000a). Danger in the field: Risk and ethics in social research. London: Routledge.

Lee-Treweek, Geraldine & Linkogle, Stephanie (2000b). Putting danger in the frame. In Geraldine Lee-Treweek & Stephanie Linkogle (Eds.), Danger in the field: Risk and ethics in social research (pp.8-25). London: Routledge.

Lisiak, Agata (2015). Fieldwork and fashion: Gendered and classed performances in research sites. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 16(2), Art. 14, http://dx.doi.org/10.17169/fqs-16.2.2334 [Accessed: October 10, 2017].

Maguire, Edward R.; Willis, Julia A.; Snipes, Jeffrey & Gantley, Megan (2008). Spatial concentrations of violence in Trinidad and Tobago. Caribbean Journal of Criminology and Public Safety, 13, 48-92.

Marpsat, Maryse & Razafindratsima, Nicolas (2010). Survey methods for hard-to-reach populations: Introduction to the special issue. Methodological Innovations Online, 5(2), 3-16.

Meuser, Michael & Nagel, Ulrike (1991). ExpertInneninterviews – vielfach erprobt, wenig bedacht: ein Beitrag zur qualitativen Methodendiskussion. In Detlef Garz & Klaus Kraimer (Eds.), Qualitativ-empirische Sozialforschung: Konzepte, Methoden, Analysen (pp.441-471). Opladen: Westdeutscher Verlag.

Moser, Caroline (2004). Urban violence and insecurity: An introductory roadmap. Environment and Urbanization, 16(2), 3-16.

Nilan, Pamela (2002). "Dangerous fieldwork" re-examined: The question of researcher subject position. Qualitative Research, 2(3), 363-386.

Nordås, Ragnhild & Davenport, Christian (2013). Fight the youth: Youth bulges and state repression. American Journal of Political Science, 57(4), 926-940.

Ortbals, Candice D. & Rincker, Meg E. (2009). Fieldwork, identities, and intersectionality: Negotiating gender, race, class, religion, nationality, and age in the research field abroad: Editors' introduction. PS: Political Science & Politics, 42(2), 287-290.

Palmer, Catherine & Thompson, Kirrilly (2010). Everyday risks and professional dilemmas: Fieldwork with alcohol-based (sporting) subcultures. Qualitative Research, 10(4), 421-440.

Przeworski, Adam & Teune, Henry (1970). The logic of comparative social inquiry. New York: Wiley-Interscience.

Rawlinson, Patricia (2008 [2000]). Mission impossible? Researching organized crime. In Roy D. King & Emma Wincup (Eds.), Doing research on crime and justice (pp.264-291). Oxford: Oxford University Press.

Sampson, Helen & Thomas, Michelle (2003). Lone researchers at sea: Gender, risk and responsibility. Qualitative Research, 3(2), 165-189.

Seawright, Jason & Gerring, John (2008). Case selection techniques in case study research: A menu of qualitative and quantitative options. Political Research Quarterly, 61(2), 294-308.

Seepersad, Randy & Bissessar, Ann Marie (2013). Gangs in the Caribbean. Newcastle: Cambridge Scholars Publishing.

Warren, Carol A.B. & Rasmussen, Paul K. (1977). Sex and gender in field research. Urban Life, 6(3), 349-369.

Westermarland, Louise (2000). Taking the flak. operational policing, fear and violence. In Geraldine Lee-Treweek & Stephanie Linkogle (Eds.), Danger in the field: risk and ethics in social research (pp.26-42). London: Routledge.

Wolf, Sonja (2012). Mara salvatrucha: The most dangerous street gang in the Americas?. Latin American Politics and Society, 54(1), 65-99.

Dr. Janina PAWELZ is a researcher at the Institute for Peace Research and Security Policy at the University of Hamburg (IFSH) and an associate research fellow at the GIGA Institute of Asian Studies (IAS), Germany. Her main areas of research include urban violence, youth violence, security and radicalization.

Institute for Peace Research and Security Studies at the University of Hamburg (IFSH)
Beim Schlump 83
20144 Hamburg, Germany

Pawelz, Janina (2018). Researching Gangs: How to Reach Hard-to-Reach Populations and Negotiate Tricky Issues in the Field [47 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 19(1), Art. 2, http://dx.doi.org/10.17169/fqs-19.1.2878.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License