An insider looking in or an outsider wannabee? Studying vulnerable hard-to-reach populations in the field of victimology – the example of the Roma communities in Sweden

Introduction

The world is getting more diverse. Post-war Sweden, in particular, has been a country characterized by increasing ethnic diversity. Since 1980 and onwards, the dominant immigration category has also changed from European labor migrants to asylum migration and family reunifications from ethnic groups outside Europe. Currently, approximately 23.5% of the Swedish population are first- and/or second-generation immigrants (Central Bureau of Statistics [SBC], 2020). In such a society, researchers must be equipped to access populations that vary along many social dimensions, including social class and ethnicity, and who have different historical trajectories and experiences, such as different migration processes, or have a background among historically marginalized minority populations. This article will focus on the study of such vulnerable hard-to-reach populations in relation to victimization research.

We can define a vulnerable or hard-to-reach population as a community of people who traditionally, and for various reasons, have been underserved by research, are exposed to more (or more severe forms of) victimization and who, due to their vulnerability, are less able to cope with, resist and recover from the impacts of victimization. ¹ Another characteristic that distinguishes such groups from non-vulnerable populations is that they are often marginalized, lack social and human capital, and are easy to ignore in the political debate. Gathering information on victimization patterns is essential so that prevention strategies and victim-supportive initiatives can be targeted adequately (Pfeffer and Cuveas, 2016). The examination of vulnerable hard-to-reach populations may also offer an essential contribution to the development of theory. Studies which re-test theories on underserved populations enable us to learn about the differences between groups and cultural contexts, thus giving us a more in-depth insight into the causes of social problems such as victimization. Finally, the research production may have a positive effect on the relationship between the majority society and the specific vulnerable population. To produce research that highlights different vulnerable populations’ suffering gives the majority society an insight into the suffering of the vulnerable community, sends a signal to the vulnerable community that they deserve to be protected, and can be used for empowering the said community. It is therefore regrettable to note that empirical studies of vulnerable and hard-to-reach populations are still uncommon in social science, including victimology. The reason for this is partly that the interest in studying vulnerable populations is relatively new, but also that there are several hard-to-overcome methodological barriers related to the study of these populations. Therefore, we must always strive to develop our research methodologies to improve the identification and recruitment of such vulnerable hard-to-reach populations.

This paper is divided into two parts. The first part will review methodological barriers to victimological empirical research on vulnerable hard-to-reach populations and highlight promising strategies for accessing such groups. In the paper’s second part, within the framework of critical reflexivity, I will draw on and reflect on my own experiences as a Roma (Traveler/resande) academic conducting a victimization study on individuals who belong to the Roma communities in Sweden. The hope is that this paper will give practical guidelines to researchers who are interested in studying vulnerable hard-to-reach populations, particularly concerning victimization research. I will start by providing an overview of the history of the Roma communities and Roma research from a Swedish perspective.

A history of the Swedish Roma communities and Roma research tradition

Linguistic analysis and DNA evidence attest that the ancestors of the Roma wandered out of northern India in the 11th century; scholars suspect this was due to the displacements caused by war. These ‘proto-Roma’ traveled from India into Persia, then Armenia, then Byzantium and thereafter spread, during different phases, through the European continent; during this time identities were formed (Hancock, 2002). Different states have handled their Roma populations in different manners, and the racist expression that was translated into political action varies significantly between different countries and times. However, what can be said is that the Roma communities have often been persecuted and marginalized by non-Roma society. Different scholars also highlight different hypotheses to explain why the Roma often suffered ideological persecution when they arrived in Europe (often labeled ‘anti-Gypsyism’). Hancock (2002) argues that it was mainly religiously and ethnically motivated, while others have emphasized that anti-Gypsyism started to develop as a consequence of different states, during a time of mass migration, targeting the group’s nomadic lifestyle (Petrova, 2003). Also, in Sweden, the persecution of the first Roma who immigrated to Sweden in 1512, known today as Travelers (in Swedish: Resandefolket), started soon after their arrival. However, relatively little is known of these early forms of anti-Gypsyism state measures in Sweden, and much more is known about the group’s situation during the first half of the 19th century.

Following the industrialization during the late 19th century, a debate arose about the so-called social question in many European countries. The social question referred to how different social problems, such as poverty, crime and housing, should be handled between different actors in society. In Sweden, the development of welfare states and their outreach to the population has been described as a social dimension of the nation-state; it not only anchored individuals’ life chances within the state, but also fostered specific ‘ways of life’ (Daly and Rake, 2003). The debates around the social question had a significant impact on how political debates developed concerning how the state should manage its Roma population (Montesino, 2002). During this time, often called the ‘folk-home era’ (in Swedish: folkhemstiden), several prominent Swedish thinkers argued for progressive reforms to help the working class. However, many also argued that no welfare policies could help certain citizens due to their antisociality. The Travelers or ‘tattare’ were believed to be one such population. During this time in Sweden, the Travelers were exposed to a wide range of social interventions by the Swedish state, such as tattare-mappings, inventories, racial examinations and sterilizations, to name a few. This was partly motivated by the fact that Travelers were perceived as a criminal sub-population of the Swedish underclass by social scientists and not regarded as a true ‘ethnic’ community (Montesino, 2002; Westin et al., 2014). Adam Heymowski (1969) is possibly the best-known proponent of this socioeconomic thesis, but this notion of the Travelers as a non-ethnic community can also be found, as shown by Strand (2001), among later Swedish scholars like Dr J Takman. Still, it should be acknowledged that the Travelers had been persecuted long before the ‘folk-home era’ and there still existed racial undertones in the debate during this time (Montesino, 2001, 2002). Many of the Travelers who participated in early ‘Roma’ studies during this time in Sweden were criminal convicts or psychiatric patients who were often forced to participate in studies, or individuals who were identified by the state as ‘tattare’ (Montesino, 2002; Westin et al., 2014).

Another Roma community, which referred to themselves as ‘Kalderash’, had left Romania after their emancipation from slavery and arrived in Sweden in 1864. They were, in the beginning, not as targeted by state interventions from Swedish authorities as the Travelers. Instead, the initial policy adopted by the state was to try to make this group leave Swedish territory. For a long time, this community was excluded from many Swedish institutions and had limited access to welfare and housing (Selling, 2013). Later, the idea was developed that these Roma (often called gypsies) would be assimilated if they were ‘rehabilitated’ from their antisocial culture (Montesino, 2001, 2002; Westin et al., 2014). Research conducted on the Kalderash community was often carried out by so-called gypsy experts, who were involved in various government investigations (Montesino, 2001, 2002). Many of these studies had social engineering as their primary purpose, making it difficult to draw a line between scientific and nonscientific works.

It should be noted that different homogenous groups have been grouped together due to their practices or history of nomadism (and due to different political processes) rather awkwardly under different labels in different countries. In many European countries, two broad populations within the so-called Roma communities are often identified. The first one is the Roma, who are part of a culture that is distinct from the dominant society where they live and have relatively recently emigrated from Eastern Europe to Central Europe, such as the earlier mentioned Kalderash. The second group is more similar to the Swedish resande (referred to in this paper as ‘Travelers’) and is often identified in England as Gypsy-Travelers, in Germany as Sinti and in Norway as Romanifolket/tattare. This second group often shares cultural characteristics and ethnic links with the Roma but identifies more with mainstream society and particularly their state’s geographical boundaries (Bancroft, 2005; Condon et al., 2019; The governments official investigations [SOU], 2010: 55). In Sweden, after the year 2000, the Travelers (resande), the Kalderash Roma and other newer sub-populations have been acknowledged as a national minority in Sweden under the juridical umbrella term ‘the national Roma minority’ (often simply referred to as ‘Roma’) (SOU, 2010: 55) and are thus more similar to the Council of Europe’s (2012) definition where Roma includes a wide range of groups. As one might expect, whichever definition is used, it has also been criticized by different individuals and communities that are perceived to be included in the label. In this paper, I will refer to these groups as ‘Roma communities’. The more recent research concerning Roma communities in Sweden includes different cultural but, above all, historical studies. Empirical studies on the current situation of the Roma communities are, however, still limited.

Methodological barriers to researching vulnerable populations

This section explores some of the challenges in the recruitment and participation of vulnerable hard-to-reach populations in research. The existing body of literature often makes a distinction between institutional and individual methodological barriers to research participation.

At the institutional level, a challenge when it comes to researching vulnerable or hard-to-reach populations is that traditional sampling methods often cannot be utilized. This is because researchers lack knowledge of the size of the specific population and the ability to locate them geographically (Heckathorn, 2011). In comparison with many other countries, Sweden lacks ethnic registration data, which makes it impossible to draw a generalizable sample of many ethnic community populations. As a way of gathering samples from such populations, several alternative sample methodologies have been utilized and developed, including the so-called snowball-sample, respondent-driven sample (Zhang, 2012) and organizational samples (Kelly, 2003). However, although these sample strategies have been used successfully to gather data on different populations, they are problematic due to their inherent bias. Simply put, these sample strategies cannot be used for collecting generalizable data or conducting comparative studies. One newly used sample methodology is to draw samples from various institutions where the vulnerable population is believed to be present. In the field of victimology and hate crime research, this strategy has been used successfully to draw samples from university students (Andersson, 2018). A sample obtained in this way can be used to compare different groups’ experiences and this is thus a significant step forward in the pursuit of conducting comparative research. However, the disadvantage of this sample methodology is that the selection excludes people from groups that are underrepresented in a specific institution. Neither can this sample be used to generalize the result outside the institution (Bonevski et al., 2014). Without reliable data on the total size of any given population, it is impossible to draw a representative sample of that group. In the end, a researcher who is interested in studying vulnerable hard-to-reach populations has to be pragmatic, and often accept limited generalizability of their research.

The most common individual barrier for participation in research, especially among the most marginalized in society, is due to resource constraints. Participant resource constraints affecting participation in research include finding time for conducting the data collection, travel arrangements and communication problems, to name a few (Bonevski et al., 2014). There are several steps researchers can take to help groups that have a limited amount of resources to participate in research, such as organizing the data collection in a way that benefits the participant, paying for transport and research participation, using translators and overall being mindful of such groups’ needs. However, both ethical and methodological issues may arise if the research projects are supporting their research participants in such a way that one could question if they are ‘buying’ the participants’ participation.

One of the first problems faced when researching vulnerable or hard-to-reach populations is the problem of labeling the community (Bonevski et al., 2014). The problem of labeling is central for laying the groundwork for trust, showing respect for the study participants’ self-identification, and for the researcher(s) to get a working definition of the group they are about to study. This can be problematic since there is seldom only one label that is used exclusively within a specific community. Identity is interchangeable and may vary significantly across age, culture and political affiliation and may even change during the individual’s life course (Modood, 2005). A flexible and commonly used approach to overcome the problem of labeling is to ask study participants to self-classify. However, a problem with this approach is that the self-descriptions that researchers are likely to collect from a community (where different individuals may perceive their identity in other ways than other members of their supposed ‘collective’) may give rise to other methodological problems, while research advertising for a clearly defined group and community may not reach all individuals with the attributes they want to examine. Research allowing too broad conceptions of identity risk lacking accuracy and meaning.

A related problem is labeling and capturing a group’s experiences. In the victimological literature, research demonstrates that labeling a crime ‘sexual assault’ or ‘hate crime’ does not always capture the victim’s perceived experience (Andersson, 2018; Orchowski et al., 2013). Thus, recruitment may be limited if study advertisements call for specific individuals who have been exposed to specific crimes, because these labels do not capture the victim’s perception. There are likely to be more challenges revolving around measurements in ethnically diverse contexts. This is due to the fact that the measurement used by researchers in a study should meet the requirements for cross-cultural equivalence, which broadly means that a given concept is measured (or captured in a qualitative context) equally well across different cultural groups (Harkness et al., 2003), and for conceptual equivalence, which refers to the extent to which an instrument captures the full meaning of a theoretical concept equally well in various cultures (Dana, 2000). Researchers interested in studying vulnerable or hard-to-reach populations should never take for granted that the measurements that they are using, even though they may have been tested repeatedly in earlier studies, will function as well in a new socio-cultural context and setting. However, the primary individual barrier to participation in research among vulnerable or hard-to-reach populations is the mistrust of the researcher and the research process (Bonevski et al., 2014; Pfeffer and Cuveas, 2016). Mistrust among a minority community may stem from these groups’ vulnerable position in relation to the researcher and from fear of stigmatization. Some individuals and groups may also have negative experiences. Refugees may have suffered persecution in their home countries (Garland et al., 2006) and the mistrust of indigenous minorities, such as the Roma communities, may be explained by the historical violations they have suffered at the hands of the society where they live. In the field of victimology, evaluations of minority populations’ participation in self-report questionnaires have indicated that ethnic minority populations often have higher social desirability bias than non-minority participants. This is partly due to ethnic communities’ fear of being further stigmatized by research results (Bowling, 1990), indicating that they do not believe in the benefits of participating in research, but also that some questions can be sensitive to answer in a certain cultural context and by certain crime victims (Pfeffer and Cuveas, 2016; Wallengren and Mellgren, 2015b). Building trust in a vulnerable hard-to-reach community will demand careful consideration and potentially extended efforts before the start of the research project. A common argument for a solution to overcome mistrust is that empirical studies should be conducted by ‘insiders’ who share their identity with the target community.

The insider/outsider debate

Since the 1990s, there have been discussions as to the significance of outsider/insider research positions when conducting empirical research. To provide a background to my reflections in regard to this debate, I will briefly describe some of those discussions here.

Those who argue that insiders have methodological advantages often propose that the insider status helps the insider to be more aware of the lives of the study participants and benefits the researcher in the processes of developing research questions, elaborating on research design and accessing study participants (Corbin-Dwyer and Buckle, 2009). According to these researchers, study participants are likely to give more honest and accurate responses to researchers of a similar social position, and they advocate gender, racial and ethnic matching of the researcher concerning the community targeted for a study (Essed, 1990). It is also argued that insider-researchers are better suited to represent marginalized communities compared to an outsider (Bridges, 2001).

Critics, on the other hand, have argued that research by an insider does not necessarily lead to methodological advantages or more accurate interpretations of the study participants’ experiences (Gunaratnam, 2003). These critics highlight the fact that study participants may have particularly high expectations of insider research and place unrealistic hopes on the research project (Kanuha, 2000); that there exists a risk that the breaking down of the boundaries between the researcher and study participant may cause study participants to disclose more to the researcher than they are comfortable with (Watts, 2006); that the insider-researcher might take for granted study participants’ perspectives due to their shared position; and that some issues might not be mentioned because of assumptions of shared knowledge (Perry et al., 2004). Some have argued that an outsider-researcher may even hold a methodological advantage due to their detachment, which provides objectivity (Savvides et al., 2014). It is important to acknowledge that social identities are multifaceted and complex. Sharing one aspect of participants’ identity, such as ethnicity, does not guarantee that a researcher will understand a study participant’s experiences any more than someone from another ethnic group, since other personal characteristics may outweigh the researcher’s and study participants’ similarities (Bridges, 2001). Still, it is well established in the empirical literature that study participants may give desirable answers based on the identity of the researcher who is involved in a study. In relation to the researcher’s ethnicity, this phenomenon is sometimes called the ‘race of interviewer effect’ and refers to the response bias and measurement error that have been recorded in the adjustment that study participants make to their opinions and attitudes when questioned by a researcher from another ethnic group, in interviews or when answering a questionnaire (Gunaratnam, 2003; Pompper, 2010). Krysan and Couper (2003), for example, found that African American study participants expressed greater negative feelings toward whites when the interviewer was African American compared to when the interviewer was white. Furthermore, both African American and white study participants expressed more conservative attitudes toward racial policies and modified their perception of discrimination in the dialogues with the interviewer of the opposite race. Thus, empirical evidence shows that the ethnic identity of the researcher does play a role in the research process, highlighting the importance for researchers of reflecting on how their racial or ethnic identity affects their research.

Insider perspectives at different stages of research

Before reflecting on the impact of my own insider status on the research process, it is important to contextualize this discussion by offering a brief overview of the research study that is under discussion.

Research design

It is often claimed by scholars who argue for the benefits of an insider position that insiders have methodological advantages in the early stage of a research process (Bridges, 2001; Corbin-Dwyer and Buckle, 2009; Essed, 1990). Indeed, in the planning phase of the project, I do believe that my ethnic insider position provided many methodological benefits.

In the early stage of the planning phase, Roma organizations were contacted and asked if they wanted to participate in the project, but few of these showed an interest. According to the organization’s representatives, the reasons for this were mainly due to three political events that happened before the planning phase of the study. Firstly, the nationalist party the Swedish Democrats (Sverigedemokraterna) had started to gain more ground in the political sphere in Sweden. This development scared many Roma. Secondly, the Swedish state was actively promoting programs related to Roma integration in Sweden, something which resulted in many Roma organizations lacking the resources to participate in other time-consuming projects. Thirdly, it had been found out that the Swedish police had an illegal database of Roma in the southern parts of Sweden. The Roma communities perceived this as proof that the state was willing to discriminate against them (Wigerfelt and Wigerfelt, 2015). Some organizations’ leaders feared that involvement in a project would cause discomfort in the communities and that they lacked resources to participate.

Instead, and due to my insider position, I was able to use my previously established network of contacts with individual Roma to reach potential study participants. In total, 16 gatekeepers helped with the data collection. Thus, by using the snowball-driven sample and thanks to the already existing network of individuals I had gathered during my period as a Roma activist, we were able to collect a relatively large sample of Roma to participate in the study. Most of the gatekeepers expressed that they would not have been willing to help with the study if it had been conducted by an outsider. Thus, in regard to accessing the study participants, my insider position had many benefits. I also found that my insider position helped me in the marketing of the study and with the construction of the questionnaire and interview guide. As highlighted earlier, a common problem regarding research on vulnerable populations is how the target community should be labeled and to whom the study is directed. The Roma are a heterogeneous population consisting of different ethnic groups with varying perceptions about their origin, history and current place in society and, as we have seen, they are defined differently in different countries (Brown and Scullion, 2010; Condon et al., 2019; Kósa et al., 2007). While some have a strong ethnic identity as ‘Roma’, others perceive this term only as a legal umbrella term used as a means for different groups to organize themselves politically. For yet others, the term ‘Roma’ is perceived as a discriminatory term applied to them by the majority society (and a ‘Roma elite’) as a way of stigmatizing them further and not letting their true identity become acknowledged (Brown and Scullion, 2010; Heaslip et al., 2018; Wallengren and Mellgren, 2015b). As a member of the Roma community, but more specifically as a member of the Traveler community (which is the group who most aggressively have tried to distance themselves from other Roma sub-populations; Hazell, 2011), I was often asked how I related to these questions and if I believed that some groups should be acknowledged as a separate minority. I answered these questions by explaining my own often changing beliefs about this issue. I also argued and tried to convince individuals that my study would use the legal umbrella term but, at the same time, acknowledge the wish of different groups to distance themselves from other groups. Thus, to some extent, I tried to be outside the political debate about the ‘Roma’ label. For the individuals who helped with the study, this seemed to be a sufficient answer.

Still, the problem existed of how I would define the population. According to some individuals who helped with the study, it would be impossible for us to get access to the field if we, simplistically, used the term ‘Roma’. Others suggested that it would be problematic to use other forms of labels. After discussions and arguments, we arrived at a joint conclusion. We would let respondents use their self-identifying terminology. ‘Roma’ in this project thus refers to groups or individuals who specifically refer to themselves as Roma, but also to individuals who considered themselves to be labeled ‘Roma’ by others. We later found, when we were conducting the data collection, that it was indeed essential to use different labels for different individuals and/or groups. In both the interview material and the questionnaire, it became more and more apparent that several of the study participants would not have participated if we had used only one label to describe them. The use of multiple self-describing ethnicities was also beneficial for conducting within-group differences in the material. I believe that a non-Roma researcher would have had more problems navigating around these issues of labeling the population.

Regarding the designing of research questions and measurements, it has been argued that insiders hold an advantage over outsiders. Insiders are believed to be more familiar with the types of issues that affect the marginalized community and, therefore, better suited to develop and be aware of what types of questions should (or should not) be included (LaBare, 2002). In the project, the knowledge of the Roma that I had developed, combined with insider knowledge about different cultural taboos, was beneficial when constructing the questionnaire and interview guide. However, due to the heterogeneous nature of the Roma, I found it necessary to engage in an iterative process, where I asked the gatekeepers for feedback on the questionnaire. I was surprised that there were, especially among some Roma sub-populations, culturally distinct from my sub-group, other social taboos that I had not thought would be of importance. Thus, I found the discussion process with the gatekeepers to be very useful, as I realized that there could be issues with the study that I had not thought of or believed were of importance. Due to the discussion with the gatekeepers and by reviewing the earlier literature about possibly sensitive subjects within the Roma communities (Goldstein-Kyaga, 1990; Weyrauch and Bell, 1997), I concluded that it would be better to exclude some questions from the survey.

The gatekeepers highlighted that several background measurements in the questionnaire were too sensitive to gather – in some cases due to different cultural beliefs and in other instances due to fear of stigmatization. Firstly, the gatekeepers were critical of the inclusion in the survey of measurements of socioeconomic status (including income, education level or type of work). The reason for this was twofold. Some of the gatekeepers were concerned about the stigmatization that the analysis of such variables might cause the communities. As we have seen, from a historical perspective, the Roma have become vulnerable because of their alternative means of supporting themselves economically (Selling, 2013). Some gatekeepers told me that they and other Roma would not like to highlight their marginalization further. For other individuals, they argued, the reason was that some Roma were thought to be unwilling to ‘show off’ their marginalization to other Roma individuals. Thus, for them, it would possibly have been better if a non-Roma had gathered and analyzed the data. Other measurements that different gatekeepers advised me not to include were where they lived geographically (due to fear of being identified and/or registered), their civil status or whether they had children. The reason for this is that the Roma culture is so family-oriented that they thought it might be difficult for a person to admit that he or she did not have a partner or children. Some also argued that in some Roma cultures it is taboo to ask someone if they have children because it signals that they have had sexual intercourse (which is a topic of conversation to avoid) (Goldstein-Kyaga, 1990; Weyrauch and Bell, 1997). Others did not want to include the question about children because there was the belief that this cannot be asked due to the fact that older Roma individuals may have been forcibly sterilized.

Researchers often highlight the need to have an intersectional perspective on victimization concerning victims. The term ‘intersectionality’ was coined by Crenshaw (1989), arguing that constructed notions of ‘difference’ impact a particular group simultaneously, such as race, sex and social class, in a cumulative manner. This perspective was particularly important for my study since one of the goals was to analyze the consequences of victimization. However, I was told by the gatekeepers that it would be problematic to use such a perspective in regard to the Roma communities. First of all, as with regard to certain background variables, the groups feared the stigma of highlighting specific social problems within the communities, including, but not limited to, their involvement in criminal activities and the prevalence of mental health problems. Secondly, as I was told, the Roma are, in general, less tolerant of some social problems within their communities than non-Roma. This is likely to be due to the Roma communities’ suffering during the 20th century being heavily motivated by the state’s wish to prevent the communities’ ‘antisociality’ (such as mental health problems, criminality, alcohol and narcotics problems) (Westin et al., 2014). The gatekeepers were also critical of incorporating questions related to particular problems faced by females and Lesbian, Gay, Bisexual, Transgender/Transsexual (LGBT) individuals. The reason for this was, mainly, that some of these groups suffer from particular problematic structures within the communities and they (and potential study participants) would not be willing to let these problems become visible to the non-Roma community.

It should be noted that this lack of measurement was only applicable to the questionnaire. In discussion with the gatekeepers it was concluded that if a particular study participant talked about any of these sensitive issues during the qualitative interview phase of the study, it would, both from an ethical and a methodological perspective, be desirable to let the participant elaborate on his or her thinking about the particular subject. During the qualitative phase of the study, several study participants highlighted a number of issues that were perceived to be too sensitive to cover in the survey and that related to the perspective of intersectionality. To summarize, I argue that I had an initial advantage in formulating research questions, and constructing and selecting measurements. However, it was still essential to ask the target community for guidance. This is particularly important for a group like the Roma, who do not have one specific culture and history and who have quite different cultural taboos within their communities. Finally, a problem I encountered after the questionnaire had been answered by the respondents was that I could conclude that while I had developed a certain measurement for the inclusion of particular Roma sub-communities, these groups only participated to a limited extent in the study (Wallengren et al., 2019; Wallengren et al., forthcoming). This raises the question of how a researcher should handle a population where different subjects can have a different stigma for different groups, and possibly only a small minority of the population do not want to include important measurements. As I was careful to meet different groups’ needs, I developed the questionnaire using ‘the smallest common denominator’, meaning that most Roma communities’ members would have been able to answer a particular question without feeling stigmatized or harmed.

Data collection

Looking back at the data collection, I do believe that I experienced some benefits in being a member of the Roma communities that a non-Roma would not have had in the same context. I could use my cultural competence, talk to study participants in their own language (or at least a language they perceived as close to their own), understand cultural issues within the communities and overall relate to ‘being a Roma’.

There were study participants who told me that they would not have participated if a non-Roma had conducted the study. However, a couple of study participants told me that they had participated in other research projects earlier but that they, in these instances, had lied to the non-Roma researchers. The reason for this, I was told, was that the participant was not willing to participate in a study, but felt forced to do so due to the need of the communities to market themselves. Some also argued that they had chosen to participate in research as a way of tricking the non-Roma researcher and ‘having fun at their expense’. For these study participants, the goal was to give the researcher incorrect and ‘absurd’ information so that later they could talk to other Roma about how easily tricked and naive the researcher was. Even though I cannot attest that the study participants were honest during our discussions, I do believe that my insider position made the study participants less willing (and possibly less able) to lie about their experiences. However, there were situations where I believe my insider status made the data collecting phase problematic.

During two interviews, I noted that the study participants seemed to ‘wink at me’ during our discussions. After the interviews, they told me that they had lied regarding some aspects of the interviews to give me interesting data material. In a couple of interviews, I also noticed that the study participants were willing to discuss political policy concerning the Roma but less willing to discuss questions related to the study. These study participants did tell me that they were motivated to participate in the study because they hoped it would contribute to reaching some political outcome (including separating themselves from other Roma sub-populations or making the Swedish government give the Roma communities compensation for their historical suffering, etc.). My interpretation is that the study participants were willing to tell me this because I belong to the community and would be willing to further these political goals myself.

The question then arises whether or not my insider status made the study participants more inclined to lie and try to further political agendas, or if my insider status made them more willing to admit to having other primary motives for participating in the study. Another situation where I found that my insider position affected my interactions with the participants negatively was in the interaction with females, since I am a Roma man. In some Roma communities, sexuality is stigmatized. If one does not follow social codes of ‘purity’, it is at the very least considered rude, but for some it is even believed that breaking such social codes can lead to sickness and other forms of negative outcomes (Weyrauch and Bell, 1997). One such behavior, for more conservative Roma, which is often seen as problematic is for a man and a woman to be alone together in a closed setting. In one interview, a particular woman’s family member insisted on ‘inspecting’ our conversations (see Bhopal, 2010 for an interesting discussion about gender identity and researching marginalized populations).

A related problem was my relative youth. Within the Roma culture, there are norms on how a younger individual should behave towards ‘elders’. This includes, in some communities, using language that signals the status of the older individual, not talking about certain issues (primarily related to sexuality) and not ‘questioning’ the elder’s statements. In most instances, I do believe that I succeeded in orienting myself in these contexts due to my cultural insider position. However, I sometimes found it difficult since I do not uphold such norms in my personal life. I also want to highlight that my insider position can make the interaction with Roma study participants, at least in some instances, more difficult; I would be likely not to meet these problems if I was considered a non-Roma, since non-Roma are excluded from these norm systems (Weyrauch and Bell, 1997). I believe that these rules, at least sometimes, are only upheld because Roma, when interacting with one another, do not want to be the first one to break the rules even though they do not consider them to be important.

There were other times when, though I was still seen as an insider, certain aspects of my identity were perceived as outsider attributes. First of all, one specific regard in which my insider status was problematic was that, due to the fact that I had worked with Roma issues before the research project started, there existed statements that I had made in different aspects of political life. I was sometimes asked why I had said or done something specific in different political forums. One example was my involvement in the PRIDE parade and a lecture that I had given during the event about similarities between living with different stigmas, comparing research revolving around being Roma and HBTQ individuals. Several of the study participants that I talked to were critical that I had mentioned the two groups in the same political context. As a Roma, I was told that I ‘should know better’ than to talk about such immoral behavior. Some of the study participants also questioned the fact that I was unmarried and childless. Due to my involvement in the PRIDE parade, being single and not having any children, some individuals involved in the study asked me if I was gay and told me that if I was, they would not like to participate in the study. Still, there were study participants that I met who were positive about my ‘liberal’ values and told me that they, being gay, bisexual or having a ‘norm-breaking lifestyle’, would not have participated in the study if they had not known about my political leanings and progressive attitudes.

My insider position was sometimes rejected by individuals involved in the study. Since I do not have visible characteristics that signal group belonging, I sometimes felt that I had to highlight my identity in other ways in face-to-face interactions, such as during interviews. I did this by emphasizing my language skills, giving study participants an insight into my family history and actively dressing in a more ‘Roma way’. Sometimes I found it challenging to make all of these behavior changes and highlight my group identity. Especially during interview situations, I felt interviewed by the study participants and ‘checked’ as to whether or not I was indeed a Roma. In the questionnaire, it was, to some extent, possible to avoid study participants questioning my group belonging due to their lack of ability to perceive identity markers. There were also study participants who rejected my ethnic insider position. This was especially common among Roma individuals who belonged to other Roma sub-populations than my own. One of my gatekeepers told me before I was to conduct an interview with a particular study participant from the other Roma sub-population that I should downplay my Roma identity so as not to be perceived as silly or foolish. When I met this study participant, I was told that neither the larger group of Roma to which I belonged nor I personally was perceived as a ‘true Roma’ but as a ‘wannabee’. The reason that such study participants participated was not my ethnicity, they argued, but the importance of the research subject and that I, as an individual but not as a member of their ethnic group, was to be trusted. For this reason, it is difficult to distinguish between the methodological benefits I got during the research process due to my ethnic insider position and the trust I had from the communities as a consequence of working politically with Roma issues, as well as the ‘borrowed’ trust I got from the gatekeepers involved in the study.

Data analysis

While many researchers argue that an insider-researcher may benefit from their position when analyzing data, others have been critical and questioned the insider’s bias and ability to be objective (Kanuha, 2000). I do think that this is a valid criticism. However, I also argue that it is naive to think that bias can be fully eliminated by either insider- or outsider-researchers.

I took several steps to ensure that the interpretation of the data was as objective as possible. As already mentioned, in this study, a sequential mixed-methodology design was applied, in that I wanted to obtain qualitative data in order to explain or expand quantitative results and vice versa, thus providing a better understanding of the Roma communities’ victimization experiences. While mixed methodologies have often been highlighted as beneficial in strengthening a study’s validity and letting different data sets inform one another (Creswell and Plano Clark, 2007), I would also argue that mixed methodologies let a researcher check for different biases during the integration phase of the analysis. Different data sets gathered with a mixed-methodology design can either confirm the other’s results, expand on the others by enabling insights into the phenomenon of interest in describing complementary aspects of that phenomenon, or demonstrate discordance if the qualitative and quantitative findings are inconsistent. When analyzing the data in this study, the different data sets typically confirmed and expanded on each other, minimizing the risk of bias. I also checked my interpretations with the study participants, provided a detailed description of both the setting and the study participants involved in the study, and purposely used a so-called ‘stage approach’ when reporting my results, which means that I reported the quantitative and qualitative data separately so that both the study participants and the readers could determine the transferability, accuracy and validity of the findings.

I also want to highlight that during analysis it is not necessarily appropriate to simply be a conduit for participants’ stories and that at times it may be necessary to challenge study participants’ accounts. I found that my Roma identity sometimes helped both me and study participants to think critically about the research data. For example, some participants reported that their marginalization was due to the communities currently and historically being discriminated against by non-Roma society. I think that I had the ability to challenge participants during interviews, as I was not afraid to offend and was ‘allowed to’ question why, then, other groups with a similar history to the Roma (e.g. the Jewish minority) are not, today, as marginalized. I believe it may be easier and possibly more ‘accepted’ for an insider to challenge study participants’ accounts.

Lastly, I think that something must be said about conducting research on a population that one identifies with. During the research process, I heard stories of people telling me about their experiences of being exposed to hate crimes, bullied and raped. I listened to parents telling me stories about their children committing suicide and people living their whole lives with feelings of inferiority. Even though I think that such stories may affect all researchers of marginalized populations, I think it is likely that insiders are often more negatively affected than outsider researchers, and I would advise insiders to take precautions. In addition, among scholars interested in Roma communities (Beebeejaun et al., 2014; Mirga-Kruszelnicka, 2018), it has been argued that an insider-researcher’s engagement in knowledge production is essential for promoting diversified, balanced and context-sensitive discourses. I certainly agree that diversity is important within academia, especially among such populations that are understudied. However, I would not want to discourage people from conducting research (as insiders or outsiders) concerning vulnerable hard-to-reach populations. The field is new and exciting, and it is important for the well-being and protection of society’s most vulnerable. No single individual or community can expect to solve the many problems in human existence – all of us are needed.

Conclusion

Across the world, the challenging ethnic diversity in regard to research has become more important. Diversity raises both theoretical and methodological questions that are not easily resolved, and requires well-thought-out study adaptions at each stage of the research process, from the development of a study to data analysis. As this article illustrates, making an effort to include vulnerable hard-to-reach populations in victimological research is important in order to obtain more ethical, valid and useful research findings. Furthermore, as we saw from the earlier literature, a possible way of both accessing vulnerable populations in research and increasing research project success is to use ‘insiders’ to study the target population. An overall conclusion from my own experience in this matter, and something I hope has been apparent in this article, is that conducting research on different underserved and marginalized populations is not only important, but also possible. As a self-proclaimed insider looking in, I will conclude with the statement that so-called ‘insider’ research has its benefits. Still, it is difficult to truly conclude who an ‘insider’ is, especially since research subjects do not agree on what characteristics should be considered ‘insider’ attributes. During my research project, I was often considered a group member, an insider. This gave me some methodological benefits, but also some problems that I otherwise would not have had to deal with. For some, I was not seen as an insider but rather an outsider-wannabee. After reflecting on these issues, I consider myself to be something ‘in between’ the insider and outsider – a position I believe that many researchers find themselves in when studying a community whose ethnic group belonging they share.