Trust Toward the Criminal Justice System Among Swedish Roma: A Mixed-Methodology Approach

Minority populations’ trust toward the criminal justice system is understudied in many parts of Europe, including Sweden. This article will contribute to this field by examining the trust in the criminal justice system among the Roma community in Sweden. The aim of the study was to (1) estimate the Roma community’s trust toward the criminal justice system, (2) examine what factors influence the community’s trust toward the criminal justice system, and (3) analyze whether trust toward the authorities influences the Roma community’s willingness to report victimization. The study used a mixed-methodology design in combining survey data (n = 610) with in-depth interviews (N = 30). The findings show that the respondents have a low level of trust in the criminal justice system authorities. According to the regression analysis, the strongest predictor of trust was shown to be explained by the respondent’s perception of procedural unfairness. Qualitative findings supported these results while also highlighting cultural effects and historical processes that explain the community’s lack of trust. Finally, trust in the authorities seems to be an important factor that influences crime reporting.

justice system is often significantly lower among specific ethnic minority populations than among nonminority citizens (Bradford et al., 2015;Weitzer & Tuch, 2005). Minorities' trust in the authorities is understudied in many parts of Europe, including Sweden, where most studies on the trust in law enforcement agencies tend to be based solely on the general Swedish population (Van Damme et al., 2015). This study will contribute to the research field by analyzing a hitherto unexplored ethnic community in Sweden, namely, the Roma and their trust toward the criminal justice system.
The Roma are a transnational minority of culturally distinctive ethnic subpopulations that are characterized by their history of marginalization. There are many reasons to gather estimates of such a population's trust toward the authorities. If empirical examinations indicate that the trust of a specific population in the criminal justice system is low, this could be interpreted as a severe social problem. There is a risk that criminal investigations will be difficult to conduct if the authorities do not enjoy the legitimacy and trust of the community they are trying to support (Tyler & Fagan, 2008). If a specific community does not report the crimes they suffer, there is also a risk that the reporting statistics will be distorted, which may have negative effects on the legal system's resource allocation (Skogan, 2009). From a victim's perspective, low trust in the authorities may lead to feelings of exclusion and reduced levels of security, which in turn may make these groups further vulnerable (Bradford & Jackson, 2016). However, since Sweden lacks ethnic register data, policymakers are unable to draw any conclusions about many communities that may be in greatest need of support. Another reason to examine the Roma community's trust in the authorities is that the empirical examination of underserved populations may offer an important theoretical contribution by enabling us to learn about the differences between groups and cultural contexts and thus give us a more in-depth insight into the development of trust.
For this study, a questionnaire was distributed, and in-depth interviews were conducted among Roma living in Gothenburg and Malmö (Sweden) to answer the following questions: To what extent do the Roma trust the criminal justice system? What determines the Roma community's trust in the criminal justice system? and Finally, does trust toward the authorities influence the Roma community's willingness to report victimization? This study thus has the potential to expand our understanding of ethnic minority populations regarding law enforcement.

The Roma
It is believed that the ancestors of the Roma, for unknown reasons, wandered out of Northern India in the 11th century toward Europe. In Byzantium, this "proto-Roma" culture matured to an "authentic" Roma culture with Greek cultural influences. Soon afterward, the Roma spread out across the whole European continent. The first Roma who arrived in Sweden, in 1512, are today known as "Travelers" (in Swedish "Resandefolket"). The organized persecution of the Roma community (often labeled anti-Gypsyism) started, as for other Roma groups in other European countries, relatively soon after their arrival in Sweden (Montesino, 2002;Westin et al., 2014).
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 by different actors in society. In Sweden, the debates around the social question also had a substantial impact on how the Swedish state acted against the Roma. During this time, the Travelers were exposed to a wide range of human rights violations by the Swedish state, including forced sterilizations and the forced removal of Traveler children. Another Roma subgroup, often called "Swedish Roma" (referring to themselves as "Kalderash"), who had left Romania after their emancipation from slavery and had arrived in Sweden in 1864, were in the beginning not as targeted by state interventions by 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 period, this community was excluded from and had limited access to welfare and housing (Montesino, 2002;Westin et al., 2014). After 1999, the Travelers, the Swedish Roma, and other Roma subpopulations, such as the Finnish Roma (who arrived in Sweden in the late 1950s), the so-called non-Nordic Roma (who arrived in Sweden from different Eastern European states after the fall of communism in the 1980s), and newly arrived Roma (who are the Roma who arrive in Sweden today), are all acknowledged as a national minority in Sweden under the juridical umbrella term "Roma" (Statens offentliga utredningar [SOU] 2010:55). However, the Roma are best to be regarded as a transnational minority of culturally distinctive ethnic subpopulations. Empirical examinations of the current situation of the community are limited in Sweden; however, it has been argued that the group's historical suffering has contributed to a situation where the Roma have ended up in social, political, and economic exclusion as well as contributing to the low confidence that Roma exhibit toward the majority society (SOU 2010:55).

Previous Research and the Study's Theoretical Point of Departure
Thus, the citizens' perception of the legitimacy of, and their trust in, the criminal justice system is important. However, the conceptualization of legitimacy and trust is debated. One could consider legitimacy as the belief that institutions have the right to command respect and that their use of power is justified, while trust, in contrast, depends on the evaluations and expectations regarding the behavior of the authorities (Huq et al., 2017). Still, the concepts overlap, since the trust citizens feel for the authorities lays the foundation for the justification of the criminal justice authorities' use of power and will, in this way, affect citizens' perception of the authorities' legitimacy. To explain the emergence and maintenance of trust, several frameworks highlighting both cultural socialization and rational evaluation, based on direct (and indirect) experiences of the authorities, have been developed.
Within the criminological literature, the most common way of analyzing citizens' trust in the criminal justice system has been informed by the instrumental model and the procedural justice model. Although these theories emphasize different social mechanisms that are believed to create trust in the criminal justice system, there are several overlaps between them. Most importantly, both of these models emphasize that individuals and groups are rational actors and that direct and indirect experiences are the main mechanisms that generate trust toward the authorities.
According to the instrumental model, the criminal justice system's ability to meet citizens' expectations in combating different forms of social problems, such as crime and unsafety, is central for their evaluation of the authorities and the development of trust (Skogan, 2009). Thus, this model is "outcome-oriented." Several empirical studies have shown that successful police work is indeed related to positive attitudes toward the authorities (Kampen & Walle, 2006;Skogan, 2009;Van Damme et al., 2015). According to this perspective, ethnic minorities' lack of trust toward the criminal justice system authorities can be explained by these groups' vulnerability. They are more victimized than nonminority citizens, and they tend to live in communities with high levels of criminal activity, a situation which contributes to their low trust toward the police by sending signals that the police are unable to protect them they come to expect disorder to be a part of their lives and develop mistrust toward the authorities sometimes referred to as "legal cynicism" (Anderson, 2000;Kirk & Papachristos, 2011). The procedural justice model is primarily concerned with the importance of procedural fairness, a term that relates to notions of respect, equality, and neutrality with regard to individuals interacting with authorities (Sunshine & Tyler, 2003). Thus, the procedural justice model highlights that citizens are concerned with procedures independently of their effects on outcomes and that fair procedures are often valued more than favorable outcomes (Tyler, 2005). Research has indicated that in comparison to majority citizens, minority citizens often perceive their interactions with law enforcement negatively (Tyler, 2001(Tyler, , 2005Weitzer & Tuch, 2005). This pattern of perceived negative encounters with the authorities has also been shown for the Roma community in both Swedish and international studies (European Union Agency for Fundamental Rights, 2010;Wallengren & Mellgren, 2018;Wigerfelt & Wigerfelt, 2015). However, in international research, there are mixed views about whether or not the authorities are actively discriminating against ethnic minority populations (Anwar & Fang, 2006;Fryer, 2015). In Sweden, too few studies using valid study designs, where minority status has been isolated from other factors, have been conducted to draw any meaningful conclusions. Still, it should be acknowledged that empirical evidence suggests that it is easier to damage trust than to build it between minority members and the authorities (Skogan, 2009). Also, there may be a tendency for minority group members to share their negative experiences of the authorities, and this may affect the overall perception of the authorities negatively (Buckler et al., 2011). It is therefore possible, even though it is relatively uncommon for minority populations to be directly discriminated against by authorities, that one single negative incident can have profound negative effects on a whole community. One heavily media-covered incident from Sweden concerning the Roma community happened in the autumn of 2013, when it was revealed that the police in Skåne had an illegal database consisting of 4,000 Roma individuals. Likely, this incident has severely damaged the Roma community's trust in the Swedish police (Wigerfelt & Wigerfelt, 2015).
The influence of indirect or vicarious procedural justice judgments (i.e., individuals' judgments about how other people are treated by actors within the criminal justice system) has recently received growing interest (Kaiser & Reisig, 2019). From a socialization perspective, it is argued that the legal socialization process starts in childhood and is influenced by direct contact with noncriminal justice authority figures (including parents and teachers) and friends' vicarious experiences. Although research has yet to sort out the different components of legal socialization through early adulthood, research shows that family and friends play a role in the formation of legal judgments and trust (Wolfe et al., 2016). It is also possible that ethnic differences in socialization styles exist and that minorities are more affected by vicarious experiences than nonminorities. Research has indicated that minority parents often try to equip their children with the tool kits they need to develop resilience to discrimination and the ability to interact in a racialized social system in which one can expect bias from different actors within that society, including the criminal justice system authorities. In essence, the children of minorities are often taught during the socialization process that the world is a dangerous place, where not everyone is treated fairly (Hughes et al., 2006;Unnever & Gabbidon, 2011).
To summarize, the specific source of information from which trust and procedural justice judgments are made include both personal and general judgments based on direct and indirect experiences as well as indirect, vicarious experiences. This current study will contribute to the research field by examining the Roma community in Sweden and its trust in the criminal justice system.

Method and Data
The lack of ethnic-based registration data in Sweden, coupled with the Roma community's low trust toward the non-Roma society, makes it difficult to get research access to the community. It was, therefore, essential to find an appropriate and effective way to engage Roma individuals to participate in the study. Within this current research project, one of the authors (S.W.) belonged to the Roma minority and earlier had a role as a Roma spokesperson. For this reason, previously established contacts with individual Roma could be used to reach potential study participants. Through these contacts, 16 gatekeepers were recruited to help with the data collection and access to the Roma community. In this study, the mixed-methods design was used, that is, a combination of gathering and analyzing quantitative survey data and conducting qualitative in-depth interviews as a means of exploring the Roma community's trust and attitudes toward the authorities. The data were collected in a two-phase process.

Questionnaire
During the first phase, the gatekeepers received a number of questionnaires. They were instructed to give the questionnaire to the persons whom they considered suitable for participation in the study. The inclusion criteria were that the study participant should self-identify with the Roma community, be living in Gothenburg or Malmö, and be over the age of 16 years. The distribution took place between January 2015 and October 2016. The respondents were given two options for returning the questionnaires: either to return them to the person who had distributed them or to send them by mail to the research group in prepaid envelopes. In total, 750 questionnaires were distributed, of which 610 were answered, giving a response rate of 80%. The reason that the selection area was limited to Roma residing in Gothenburg and Malmö is that these cities are believed to have the largest Roma population in Sweden. Even though the estimates concerning how many Roma are living in Gothenburg and Malmö are unreliable, it has been estimated that between 5,000 and 10,000 Roma live in Malmö and approximately 5,000-7,000 in Gothenburg (SOU 2010:55).

Study Measures
Dependent variables. Trust in the criminal justice system has predominately been tested on the criminal justice systems as a whole and the police institution separately, in earlier studies (Bradford et al., 2017). However, the dependent variables used in the current regression analysis are trust in the criminal justice system as a whole and in the police, the prosecutors, the courts, and the prison system. The participants' response was measured on an ordinal scale ranging from very low (0) to very high (4).

Independent variables.
There were four independent variables employed in this study, based on earlier research and current theoretical development. First, following the procedural justice model, individuals who have a perception of being at risk of unfair treatment by the authorities should report a lower level of trust (Sunshine & Tyler, 2003). To assess perceptions about procedural unfairness, we presented the following question to the respondents: "Do you think that the police, in general, treat Roma unfairly compared to non-Roma?" To this question, the respondents could answer no (0) and yes (1).
In accordance with the instrumental model, being a victim of a crime and feeling unsafe are detrimental to individuals' trust toward the authorities (Skogan, 2009). To measure instrumental performance, we asked the respondents about their victimization experiences, and the question to the respondent was: "Have you experienced any crime during the last 12 months?" The respondents could answer no (0) and yes (1). The respondents were also asked whether they had reported the incident to the police (0 ¼ no, 1 ¼ yes) . In an open-ended follow-up question, respondents were asked to elaborate on why they had not reported their experience to the police. With regard to fear of crime, we asked how often the respondents were afraid of being subjected to the following acts: verbal or written threats, harassment or bullying, being followed or chased, minor assault, assault, and aggravated assault. Fear of crime was measured on an ordinal scale ranging from never (0) to regularly (3). These variables were transformed into an index ranging from 0 to 21 and had a Cronbach's a value of .910. In addition, since earlier research has consistently shown that one of the most important predictors of trust in the authorities is people's feelings of unsafety in their living areas (Brown & Benedict, 2002), we also included neighborhood unsafety in the analysis. We asked how often respondents felt unsafe in their neighborhood (daytime or evenings/nighttime). Also, neighborhood unsafety was measured on an ordinal scale ranging from never (¼0) to regularly (¼3) and was transformed into an index ranging from 0 to 6 and had a Cronbach's a value of .856, indicating a strong internal consistency. The correlation between fear of crime and neighborhood unsafety had a variance inflation factor value of 2.066, indicating that there is no multicollinearity problem between the two series indexes.
Background variables. The background variables included in the questionnaire are gender (0 ¼ male, 1 ¼ female), age (0 ¼ 16-19 years, 1 ¼ 20-29 years, 2 ¼ 30-39 years, 3 ¼ 40-49 years, and 4 ¼ over 50 years of age), city of residence (0 ¼ Malmö, 1 ¼ Göteborg), resident owner (0 ¼ no, 1 ¼ yes), and subgroup membership (0 ¼ Roma, 1 ¼ Traveler). In the upcoming analysis, we combine Swedish, Finnish, non-Nordic, and newly arrived Roma to form one category Roma. The reason for combining these subpopulations into one category is that they seem to be more socioeconomically marginalized than the Travelers (Wallengren & Mellgren, 2015b) and also tend to be newer arrivals in Sweden (first-or second-generation immigrants), which, based on previous research, may be an important distinction between the two groups due to their different historical experiences and possible socialization processes (Uslaner, 2008). Even though it is difficult to predict, we hypothesize that Roma will be less likely to trust the criminal justice system than the Travelers.
In terms of demographics, about half of the respondents in this study were male (48.7%), and on average, the respondents were young and middle-aged (21-40 years old). It is not possible to compare the sample characteristics to the general Roma population since there is no ethnic registration in Sweden. However, when comparing the sample characteristics of this study with the general Swedish population, the respondents are somewhat younger than the general Swedish population. Furthermore, the sample has an overrepresentation of the subpopulation Traveler (69.8%). The entire Roma community is estimated to have approximately 50,000 members in Sweden, and about half of them are believed to belong to the Traveler community (SOU 2010:55; see Table 1).

Interviews
In the second phase of the data collection, those who had answered the questionnaire were asked whether they would agree to participate in in-depth interviews. Within the project as a whole, 52 Roma individuals who had answered the questionnaire contacted the research team. A selection of study participants was made, where the research group had the objective of obtaining a variety of individuals regarding Roma subgroup belonging, age, and gender. In the end, 30 Roma individuals were interviewed. Seventeen of the interviewed study participants were men, and 13 were women. The age of the interviewed participants ranged from 19 to 72. The interview material, like the questionnaire, contains information and questions about the study participants' trust in and perceptions of the legal authorities and about their reasons for refraining from crime reporting. Interviews took place at a location of the study participants' choosing, often the study participants' home.

Ethical Considerations
This study has undergone ethical vetting. It has been approved by the regional ethical review board in Lund, Sweden, and conducted in accordance with the recommendations for ethical research.
The anonymity requirement was heavily emphasized by the Roma community in the planning phase of this study. An advantage of this study approach, that is, of first contacting gatekeepers who then distribute questionnaires to study participants, is that no one, not even the research team, can find out who participated in the first phase of the study. However, this approach gives the gatekeepers considerable influence both over the selection process and in the treatment of the study population (Heckathorn, 2011). It was, therefore, crucial for us to highlight for the gatekeepers that they should draw attention to the purpose of the study and that participation was voluntary. The questionnaire was constructed based on earlier measurements commonly used in victimization surveys. However, based on previous research on measures that can be perceived as culturally sensitive within the Roma community (Wallengren & Mellgren, 2015a) and on discussions with the study's gatekeepers, it was concluded that some questions were not suitable to include in the questionnaire. This included measures related to respondents' background characteristics such as respondents' socioeconomic status, living situation, and education level. One important factor contributing to the group's unwillingness to include such variables in the questionnaire was that it was revealed that the Swedish police, as earlier mentioned, had an illegal database of Roma. Several of the gatekeepers asserted that they would not hand out a questionnaire that could stigmatize the group further or cause discomfort within the community. Thus, we excluded such measurements from the survey in the hope that, even though it undoubtedly affected the study's validity, it increased study participants' feelings of safety. Another question was how the incorporation of the Roma language could be used in the study. Both from a methodological and an ethical perspective, it is often argued that it is crucial to translate texts into ethnic study participants' language. However, after discussions with the gatekeepers, it was concluded that it would be difficult to translate the questionnaire into Romani chib. Instead, the gatekeepers were instructed to offer help during the process if study participants needed it. Finally, we chose not to give any form of financial compensation for participation in the study.

Analytical Strategy
The following analytical strategy was applied to answer the study's questions. We begin with a descriptive analysis of the sample in order to gather an estimation of the extent to which the Roma trust the criminal justice system.
In order to identify factors predicting the respondents' trust in the criminal justice system as a whole and in the police, the prosecutors, the courts, and the prison system, a series of ordinal logistic regression models were conducted, given that the dependent variables were measured by ordinal scales. In the regression analysis, we were mainly able to test the contribution aspects of the instrumental and procedural justice model. Independent variables used were procedural unfairness, victimization, fear of crime, neighborhood unsafety, age, gender, residence owner, and Roma subgroup belonging (correlation matrix of the variables included in the regression analysis can be seen in Appendix Table A1). We found it necessary and beneficial to complement the quantitative analysis with qualitative in-depth interviews. This gave the study participants the ability to reflect on complex concepts that could not be included in the questionnaire for cultural and ethical reasons and that were problematic to operationalize such as how cultural socialization and the Roma community's history of prosecution may play a role in the development of trust. By letting the quantitative and qualitative data sets build on and inform each other, we were given a better insight into and a better understanding of the study participants' reasons for (dis)trusting the authorities.
The last aim of this study was to analyze whether trust toward the authorities affected the respondents' willingness to report victimization. Due to a small sample and few reporting cases in the data, we were unable to conduct a comparative analysis between those who had reported crime and those who had chosen to refrain from crime reporting. Instead, the answers to the open-ended questions in the survey were analyzed thematically. The in-depth interviews were also analyzed thematically. In line with Braun and Clarke (2006), the qualitative analysis was conducted in five steps: (1) initial read-through of the material, (2) additional reading with a focus on recurring themes, (3) initial categorization of themes, (4) further processing of the themes, and (5) final categorization.

Results
The results are presented as follows: We start by giving descriptive statistics of the respondents' trust toward the criminal justice system. This is followed by the result from the regression analysis. After that, we present qualitative data from the interview on why study participants lack trust in the authorities. Finally, we present the reasons why study participants refrain from crime reporting, as shown in the thematic analysis.

Low Trust Among the Respondents
As shown in Table 1, the respondents' trust in the criminal justice system was reported to be low. Of the 610 respondents who participated in the survey, only 9.2% reported that they had high or very high trust in the criminal justice system as a whole, and 58.2% reported having low or very low trust. The lowest trust was reported for the police. Approximately 65% of the respondents believed that the police would treat a Roma unfairly in comparison to a non-Roma. Of the 221 respondents (36%) who reported having experienced at least one form of victimization during the last 12 months, only 8% reported the incident to the police.
What Determines Roma Respondents' Trust Toward the Authorities? Table 2 contains the results of the ordinal logistic regression analysis. Across all five models, the only variable significantly related to the respondents' trust was perceived procedural unfairness. Respondents who reported that they perceived that Roma risk being treated unfairly by the police reported a lower level of trust.
In regard to respondents' trust in the criminal justice system as a whole, fear of crime, age, and Roma subgroup belonging were shown to be significant predictors, in addition to procedural unfairness. The respondents who reported higher levels of fear, the respondents who were older, and the respondents who belonged to the subgroup Roma reported a lower level of trust. The respondents' trust in the police showed similar results; however, in regard to trust in the police, Roma subgroup belonging did not show any significance. In regard to the respondents' trust in the prosecutors, the courts, and the prison system, the only variable, in addition to procedural unfairness, that was found to be a significant predictor was Roma subgroup belonging. Age was found to be a significant predictor concerning trust in the prison system, as older respondents reported a higher level of trust. Gender was shown to be a significant predictor for trust in the prosecutors and the prison system, as males reported a lower level of trust. Being a residential owner was significantly associated with trust in the prosecutors and the prison system, as respondents who owned their residence reported a lower level of trust. The models are significant and explained between 29.2%, for the courts, and 48.8%, for the police, of the dependent variable by the predictors included. Reasons for distrusting the authorities are further examined, based on the qualitative in-depth interview data, in the following section.

Study Participants' Explanations of Their Lack of Trust
The study participants' explanation of their lack of trust being in line with the quantitative analysis, their lack of trust is based on the feeling that the authorities will mistreat them (Tyler, 2001). Most of the study participants believed that the police had the ability to work efficiently to help them but that they lacked the willingness to support the Roma community. One of our interviewed study participants, a man in his 40s, told us during the interview that: "It is not that they are unable. They have no interest in helping us. For them, we are gypsies. If something happens to the Swedes, then they are willing to help because they are Swedes." Later in the interview, the same man made a statement related to the concept of legitimacy, arguing that different government actors only get involved in the lives of the Roma community when they are trying to control them: "Never to help us, always to control us. They lack respect for our culture, how we want to live, and how we want to raise our children." Only three of the interviewed participants' explanations of their low level of trust in the authorities can be linked to the performance model (Skogan, 2009) due to their narratives revolving around the inability of the police to solve crime in their living areas, which are characterized by high levels of criminal activity.
However, the interview data show that participants have quite sophisticated thoughts regarding the authorities. For example, four participants argued that although the police would be reluctant to help their community, they would be forced to support these particular study participants in case they get involved in a criminal incident because they are aware of the Swedish legislation and would not allow the police to commit any violations against them. Another interviewed study participant, a woman in her 60s, highlighted the difference between different Roma individuals' ability to deal with the police as a class issue: Those like myself who know about society and have friends and family who are well educated, we can stand up for ourselves. It is worse for those who lack such resources-[they] cannot speak with the authorities and do not know about their rights. This indicates that the study participants' perception of the extent to which they can control the authorities seems to affect their feelings of safety when interacting with the authorities.
It was common for the study participants to explain their low level of trust toward the criminal justice system as a result of past negative experiences with the authorities (11 of the 30 interviewed participants). Even though many of these negative experiences seem to have taken place relatively far back in time-and for some individuals, when they lived in other countries, before becoming Swedish citizens-they still affect study participants' perceptions of the authorities negatively. Three study participants also described relatively recent, and what they perceived as discriminatory encounters, with the police. One of these individuals, a woman in her 50s, told us about an experience with the police in which she felt that the officers who had come to her apartment after her husband had been assaulted were being discriminatory. According to the woman, the police made fun of the fact that this particular family had been attacked: They came in when my husband was on the floor and had been beaten, and the police officer said: So now the family X (the family's last name; author's note) has had a "party" again. Some persons in my husband's family are known by the police, but we have never committed crimes.
Overall, a familiar theme among the study participants is that they believe the police have a perception of the Roma as criminals.
As we have seen, cultural socialization is another mechanism that has been hypothesized as an explanation for different groups' trust levels (Uslaner, 2002). This perspective views trust as a cultural trait; decision-making "rules-of-thumb," employed by individuals in unsafe contexts where information acquisition is imperfect, have evolved through a process of natural selection determined by the relative payoffs for applying certain cultural rules. According to the study participants, their historical exposure to anti-Gypsyism, especially from state agencies, such as the police, has had a lasting effect on their cultural propensity for trust. According to the study participants, this cultural socialization process is enforced by three overlapping social mechanisms: parental socialization, storytelling traditions, and the sharing of negative and discriminatory experiences relative to the authorities.
First, in line with Weitzer and Tuch (2006), who have documented that Afro-American families tend to educate their children to act cautiously around the police in comparison to White Americans, this study's participants expressed that they have been socialized by their parents to be suspicious of state authorities. One female study participant in her 20s told us: My parents have always told us to avoid the police. I always feel uneasy when I see a police officer even though I do not have any reason to feel that way. I think this is due to my upbringing.
Older study participants also highlighted these cultural effects. One female study participant in her 60s expressed some worry that she had raised her children to be "too cautious around the authorities." However, she argued that in comparison to how her parents raised her, "it is nothing. My parents tested us by screaming 'Beddo' ['the police are coming'; author's note]. That meant that we should run away and hide." The second cultural socialization mechanism that the study participants referred to regarding trust, or the lack of trust, is in line with Uslaner (2008), who argues that storytelling is one way that trust is reproduced within groups. A male study participant in his 30s described his view of the Roma culture's perception of the police by relating to the storytelling tradition that exists in his community: "You know, all songs and stories we have about the police are either about how dangerous or how stupid they are. It is no wonder that one reacts when one sees someone wearing the uniform." Third, and more importantly, several study participants asserted that they often hear stories about how their (usually older) relatives and community members have been mistreated by the authorities, something which affects their feelings toward the authorities. It is interesting to note that two study participants told us about a woman's experience of the police making fun of the fact that her family had been attacked, when her husband had been assaulted, without knowing that this woman had also participated in the current study. This shows that negative experiences with the authorities are shared between community members and possibly indirectly affect community members' trust and judgments of the authorities. At the same time, younger study participants, especially the ones who belong to the Traveler community and claim that they are more integrated into Swedish society, argue that the times are better today and that the police are not as discriminatory against the Roma as they were a generation ago. To summarize, the interviewed study participants argue that socialization and the sharing of vicarious experiences are important factors in explaining their lack of trust toward the authorities. The next section will deal with whether or not trust toward the authorities affects participants' willingness to report victimization.

Lack of Trust Influences Crime Reporting
Studies have shown that the decision to report a crime is preceded by a cost-benefit analysis, where crime victims weigh potential costs and benefits of reporting a criminal incident (Skogan, 1984). It is often assumed that trust is one of the factors that affect crime reporting. However, empirical findings regarding the link between trust toward the authorities and crime reporting are unclear (Kääriäinen & Sirén, 2011). In the analysis of the questionnaire material, 213 open answers on why respondents had refrained from crime reporting were analyzed. In line with previous research (Skogan, 2009), the respondents gave many and often overlapping reasons for why they had refrained from reporting a crime. The most common reason was that they did not perceive the incident as being sufficiently severe for motivating a police report (94 of 213 questionnaire cases). However, one third of the respondents (71 of 213 questionnaire cases) argued that they had refrained from crime reporting due to their low trust toward the police.
Three interview study participants argued that they had not reported crimes to the police because they believed that the police would be unable to meet their needs and protect them after a crime report. These study participants maintained that the police do not prioritize the forms of criminal incidents that they had been victimized for (various types of harassments). One female study participant in her 40s, who had been harassed in the living area by her neighbors, told us that: "It is pointless for me to contact the police. I have no real proof, and if I file a report, I think that they [the neighbors; author's note] will harass us further." Most study participants, however, argue that they do not trust the willingness of the police to help them. As one male study participant in his 20s told us, when we asked why he had refrained from reporting an assault he had suffered at the hands of people who attended the same school as himself: Why waste my time, you know? They do not give a shit about me, so why would I tell them. I am only giving the scumbags who assaulted me a reason to laugh when they see that the police do not give a fuck.
In the interviews, seven study participants also highlighted that they did not want the police to have knowledge of their ethnicity. This meant that even if they could expect to get support and assistance from the police in certain instances, it would be more difficult to report an incident if the circumstances around the crime were such that the police could get information about their ethnicity-for example, if the study participant had suffered an anti-Roma-motivated hate crime. Two of the interviewed study participants also claimed that they had been stopped from reporting crimes because of their family's fear of being identified by the police as Roma. During the interview, a male study participant who had suffered a hate crime said, "I wanted to report, but my family did not want that. They thought it could lead to problems." Indeed, several of the study participants argued that a common reason that the Roma do not report victimization is because of the fear that their community will be stigmatized by the authorities and that the authorities will be given the "right" to involve themselves in the community, especially if the Roma report "within-group victimization" directed toward females and gay, bi-, and transsexual individuals, for example. These testimonies highlight heavily that the group views its ethnicity as a negative factor when interacting with the authorities and that it does not trust the authorities not to stigmatize the group. At least in some instances, individuals have to take into consideration the extended family's and community's wishes of staying anonymous in relation to the police.
Finally, it has been argued that low trust in the authorities may make marginalized populations further vulnerable to victimization (Bradford & Jackson, 2016). Two participants in this study who had experiences of intimate partner violence told us that the offender had used their lack of trust toward the police as a way of controlling them and exposing them to violence. One woman, who participated in the interview and who had relatively recently been in a relationship with a man that repeatedly assaulted her, told us during our discussion that: After I had been with him for half a year he started to beat me and, you know, force himself on me. I was so scared of him. When I finally told him that I would contact the police, he said to me that "Nobody believes in a gypsy-whore." He often used that against me.
As we can see, low trust toward the authorities seems to play an important role in explaining study participants' reasons for not reporting criminal incidents but also in making them more vulnerable in relation to offenders.

Discussion
Sweden, together with the other Nordic countries, has some of the highest levels of political trust in the world, a phenomenon that Andreasson (2017) refers to as "The Nordic Gold." Rightfully so, a high level of trust allows citizens to place confidence in one another, which benefits society in many different ways. In regard to the criminal justice system, it is believed that high trust will enable cooperation with the authorities and benefit citizens by giving them feelings of safety and security (Bradford & Jackson, 2016). This study is, to the best of our knowledge, the first more extensive examination of the Roma community in Sweden and their trust in the criminal justice system, and we believe that we arrived at several important results.
The first aim of this article was a descriptive one-to gather estimates of the Roma community's trust toward the criminal justice system. The Roma who participated in the questionnaire reported an overall low trust in the criminal justice system and its particular institutions, in particular the police. Since this study is based on a nonrandom sample, a reliable comparison cannot be made between the Roma community and the general Swedish population. However, comparing the respondents' trust with that of the general Swedish population, it is clear that the respondents' trust is notably lower (Swedish National Council for Crime Prevention, 2018). These descriptive results are in line with international quantitative estimations of the Roma community's trust toward the criminal justice system authorities outside of Scandinavia (European Union Agency for Fundamental Rights, 2010). The importance of further studying the group concerning questions related to criminal justice and trust toward the authorities is underlined by these results alone.
The second aim of this study was to analyze the factors that determine the Roma community's lack of trust in the criminal justice system. As earlier research indicates, trust is generated and developed from a variety of sources and processes. In the regression analysis, we tested for the contribution aspects of two common ways of analyzing citizens' trust in the criminal justice system-the procedural justice model (Tyler, 2001) and the instrumental model (Skogan, 2009). The result of the regression analysis indicates that the most significant predictor was made by the variable related to the procedural justice model, namely, perceived procedural unfairness. Across all five models, respondents who reported that they perceived that the Roma risk being treated unfairly by the police also reported a lower level of trust. Variables related to the instrumental model showed limited explanatory contribution. Our index for fear of crime showed significant predictive value with regard to trust in the criminal justice system as a whole and in the police institution. However, neither neighborhood unsafety nor victimization showed predictive value in any of the models. These results are consistent with the findings of earlier studies on vulnerable ethnic populations that tend to highlight procedural fairness over instrumental performance (Kääriäinen & Niemi, 2014;Van Craen & Skogan, 2014). It is interesting to note that age was found to be a significant predictor in the analysis, with younger respondents reporting a higher level of trust. This pattern is not in line with earlier studies on marginalized ethnic communities in which younger minority individuals tend to report a lower level of trust toward the authorities than older individuals (Uslaner, 2002). This could be taken as an indication that the trust toward authorities is increasing among the Roma in Sweden. In addition, Roma subgroup belongingness was shown to be a significant predictor, with Travelers generally reporting a higher level of trust than other groups. We believe that the most likely explanation is that this difference may be explained by differences between the Traveler and the Roma socioeconomic status. However, we do not rule out cultural explanations for these differences.
These patterns that we found in the regression analysis were also supported by the interviews, during which study participants argued that their low trust was not, or at least not predominantly, a consequence of the authorities' effectiveness but rather of the authorities' perceived unwillingness to support the community. The study participants who were interviewed also explained that their lack of trust resulted from a cultural socialization process that has emerged as a consequence of, and in response to, the group's historical and current exposure to racial discrimination (Uslaner, 2002). In particular, we identified three overlapping social mechanisms that we believe influence the legal socialization process and (combined with structural disadvantages) lead to legal cynicism among members of the Roma community in Sweden: parental socialization, storytelling traditions, and sharing examples of negative experiences with the authorities (Anderson, 2000;Buckler et al., 2011;Hughes et al., 2006;Kirk & Papachristos, 2011;Unnever & Gabbidon, 2011;Weitzer & Tuch, 2006). Overall, we argue that this study's results indicate that personal experiences with the authorities, social disadvantage, vicarious experiences, and socialization are important factors in explaining the Roma community's trust-or lack thereof-in the criminal justice system. However, this study cannot draw a distinction between different causal explanations, and at a theoretical level, the different theoretical frameworks overlap.
The final aim of this article was to analyze how trust influences the study participants' crime reporting. Our current knowledge about whether trust in the criminal justice system and the police affects the tendency to report crimes is unclear (Kääriäinen & Sirén, 2011). However, trust levels may be particularly relevant among the marginalized and vulnerable ethnic minority populations since there is a risk that they are in danger (or perceive the risk to be high) of being exposed to structural discrimination within the criminal justice system. According to this study's results, the lack of trust toward the police was one of the main answers given by the study participants regarding why they had refrained from crime reporting. From the qualitative data, we identified that some participants do not report victimization due to their own (or the larger community's) fear of being collectively stigmatized by the criminal justice system authorities. However, due to a small sample of quantitative data and the group's low tendency to report victimization, we were unable to conduct a valid analysis in order to compare individuals who had reported victimization to those who had chosen to refrain. Most likely, there are many factors that interact in the assessment's criteria for crime reporting (Skogan, 2009).
The two most important limitations of this study are the use of a nonrandom sample and our limited number of predictors in the regression analysis. In regard to the use of nonrandom samples, it should be acknowledged that as we lack reliable data on the Roma community, we will never be able to gather a generalizable sample. In this study, based on (unreliable) estimates of the different Roma subpopulations in Sweden, we have an overrepresentation of Travelers in the sample (SOU 2010:55). Compared to other Roma subpopulations, the Travelers have a longer history in Sweden and a stronger socioeconomic position (Wallengren et al., 2019). It is thus likely that the current study sample consists of Roma with a stronger social position than many other Roma living in Sweden. If this is the case, it is possible that the general Roma population's trust is even lower than this study indicates; however, it is difficult to assess the accuracy of this statement. The fact that we were not able to test for some relevant predictors in the regression analysis (including socioeconomic status, education, media reports, the respondents' own criminal activity, police contact, etc.) can be considered problematic. In some cases, we were not able to test for predictors (such as police contact) due to a limited number of cases in the sample. In addition, some of the measurements that we constructed in the planning phase of the study were perceived by the Roma community as too sensitive to include in the questionnaire, for cultural reasons and due to fear of stigmatization. Somewhat ironically, in order for us to be able to study the Roma community (or other hardto-reach communities), the first step will always be to develop trust from the community. We believe that the fact that we used gatekeepers, that we listened to the community's wishes, and that one of the researchers conducting this study belonged to the Roma community was an effective way of gaining and developing trust from the community and in the end helped us to engage the Roma to participate in the study. We believe that our use of a mixed methodology helped us to overcome some of the study's limitations and strengthened the study's validity by enabling the research team to capture study participants' reflections about different social problems that were perceived as too sensitive to gather during the quantitative phase of the study. The use of mixed methods also allowed us to generate a theoretical understanding of the group's situation such as how the Roma community's historical marginalization influences trust and how low trust in the authorities contributes to the study participants' vulnerability. We suggest that future research considers the use of mixed-method approaches to disentangle how ethnic groups perceive the criminal justice system.

Further Research and Policy Implications
Several strategies, often based on the instrumental and procedural justice model, have been put forward to reduce negative perceptions of the criminal justice system among minority populations. Suggestions have included better training for individuals working in the criminal justice system, community policing, and campaigns to recruit minority individuals to law enforcement (Brunson, 2007;Tyler, 2005). However, to date, such programs have not been evaluated at all in Sweden and to a very limited extent internationally. Many researchers are critical of the idea that strategies based on these frameworks will be useful in promoting trust among minority populations (Sargeant et al., 2014). There is empirical support that vulnerable ethnic communities may be negatively affected by deficient police work but not whether they will be positively affected by good police work (Worden & McLean, 2017). A large body of empirical work attributes this to institutional (experience) effects, while more recently, researchers have highlighted cultural effects. In line with Kaiser and Reisig (2019), we argue that strategies that focus on improving the attitudes of minority populations regarding the criminal justice system and that are based on the (cultural) legal socialization and the effects of vicarious experiences are of great importance in developing effective strategies. However, the essential question in enabling society to develop effective strategies is to find out what leads to trust in political institutions. If the root causes of the minority population's trust are to be understood, suitable studies must be conducted. We thus need both comparative studies and longitudinal research that follows different individuals through the life course.
Another important contribution to the research field of criminal justice and ethnic minorities would be, in a Swedish context, to gather more information on whether certain populations are indeed at risk of being discriminated against by different actors in the criminal justice system. This of course has a normative value, as it concerns every citizen's equal rights, and if authorities treat groups unfairly, it sends a signal that certain groups are unworthy of protection, help, and support, thereby most likely contributing to the lack of ontological security of these groups. Currently, we largely lack knowledge of such group disparities in Sweden. Thus, for the protection of society's most vulnerable and for the legitimacy of the criminal justice system, conducting such research is of great need.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.