Abstract
The integration of refugees is a widely debated issue, yet scholarly discussions often overlook a crucial factor: The role of host society hostility in shaping this process. Across Europe, the erosion of social norms has emboldened xenophobic discourse, fueling a surge in anti-refugee sentiments. As hostility toward refugees intensifies, a critical question emerges: How does this environment affect the lived experiences of refugees themselves? One key yet underexplored aspect of refugee experiences is social marginalization, the perception of being unappreciated, excluded, or unwelcome in society. This paper contributes to the growing literature on host society antagonism by examining how anti-refugee violence in Germany influences social marginalization among refugees. Drawing on the German Socio-Economic Panel surveys from 2016 to 2017, linked to administrative data on anti-refugee violence from the Anti Refugee Violence in Germany (ARVIG) project, the study employs multilevel modeling to assess this relationship. The findings show that refugees living in counties with higher levels of anti-refugee violence report significantly greater social marginalization. These results underscore the importance of considering the broader socio-political climate of the host society when analysing integration and assimilation outcomes.
Introduction
Over the past two decades, Europe has witnessed a dramatic intensification of anti-refugee rhetoric in the socio-political environment. While public opinion surveys indicate that general attitudes toward migrants have remained relatively stable, the erosion of social norms against xenophobia has enabled more overt and hostile public discourse (Valentim 2024). This shift is particularly evident in the increasingly negative portrayal of refugees and other migrant groups in both traditional (Esses et al. 2013; Farris and Silber Mohamed 2018) and social media (Ekman 2019; Merry 2022). Politically, the rise of far-right parties has helped mainstream such rhetoric, with even centrist parties adopting stricter immigration positions, including asylum policies to maintain electoral viability (Grande et al. 2019; Abou-Chadi and Krause 2020; Dancygier and Margalit 2020). Movements like PEGIDA in Germany and the English Defence League in the UK have further contributed to a climate of antagonism that, in some contexts, has manifested in violence against refugee communities. Taken together, native–migrant relations have thus emerged as one of the defining societal challenges of contemporary Europe, shaping not only integration outcomes but also the moral boundaries of democratic inclusion. Analysing how targeted hostility is perceived and experienced by refugees therefore speaks to broader questions of social cohesion and solidarity in diverse democratic societies.
While these developments affect many migrant groups, this paper focuses specifically on refugees, individuals who have fled conflict or persecution and are legally recognized as requiring protection. This distinction is analytically important, as refugees often face unique structural vulnerabilities and social stigma, making them particularly sensitive to hostile environments. Much of the existing literature on integration and exclusion has focussed on immigrants or minority groups more broadly, examining how the hostile socio-political climate affects migrant integration and well-being (Simonsen 2016; Bor et al. 2018; Fouka 2019; Abdelgadir and Fouka 2020; Curtis et al. 2021; Tyrberg 2021), but fewer studies have addressed how symbolic and physical violence directly shape refugees’ subjective experiences of marginalization.
This paper contributes to the growing literature by examining how exposure to anti-refugee violence affects the social marginalization of refugees. I define social marginalization as a subjective, group-based perception that one’s community is unappreciated, excluded, or regarded as inferior by the host society (Hiott et al. 2006; Bollwerk et al. 2022). This concept goes beyond individual experiences of discrimination, capturing a broader and more relational sense of collective exclusion. While it overlaps with concepts like disidentification or psychological integration, it is distinct in important ways. Disidentification focuses on individuals turning away from the national identity of the host society, whereas social marginalization captures the feeling of being pushed away, regardless of one’s desire to integrate. Similarly, while psychological integration captures feelings of closeness or belonging, marginalization taps into the felt rejection from the host society, often arising in hostile environments. This perspective matters because it allows us to understand not just how refugees behave in response to violence, but how they interpret their place in society, whether they feel included, valued, or systematically sidelined. It therefore offers a complementary and more nuanced lens on the consequences of contextual hostility, expanding beyond traditional models focused on integration or identity.
Theoretically, I argue that anti-refugee violence contributes to social marginalization by signaling group-level rejection and reinforcing a broader climate of exclusion. Such violence, often embedded within hostile political and media discourse, increases both refugees’ personal experiences of discrimination and their perception that their group is unwelcome in society. These experiences function as a cultural stressor, generating stigma, and undermining belonging in the host society. In response, refugees may withdraw from majority-group interactions and reorient their social ties inward, relying more heavily on coethnic networks. This behavioral and psychological reconfiguration driven by perceived hostility ultimately deepens their sense of marginalization.
To empirically assess these dynamics, I focus on Germany, a case that meets the key scope conditions for this study: A large refugee population and substantial subnational variation in anti-refugee violence. Between 2015 and 2019, Germany received over 1.7 million asylum applications, the highest in the EU during the 2015 migration crisis. This period also saw a significant rise in xenophobic violence, including arson attacks on asylum shelters and assaults targeting refugees. The combination of large-scale refugee arrival and variation in local hostility makes Germany a particularly suitable context for examining how anti-refugee violence shapes patterns of social marginalization.
This study employs a multilevel modeling approach to examine how anti-refugee violence impacts the social marginalization of refugees in Germany. The data for the main independent variable, anti-refugee violence, is obtained from Anti-Refugee Violence in Germany (ARVIG) dataset, a geo-referenced database documenting over 4,800 violent incidents against refugees between 2014 and 2017. Social marginalization, the dependent variable, is operationalized through three measures obtained from the German Socio-Economic Panel (SOEP) surveys, focussing on personal experiences of discrimination, feelings of being an outsider, and perceptions of feeling welcome. Using mean-centering techniques, the analysis separates within-county and between-county variations in violence.
The findings reveal that higher levels of anti-refugee violence at the county level are associated with increased social marginalization among refugees. Refugees in counties with heightened violence report greater feelings of being outsiders in German society and higher levels of personal discrimination. The final measure of social marginalization, feeling welcome, shows negative coefficients for the main independent variable; however, these results do not achieve statistical significance and narrowly miss the significance threshold with a p-value of 0.052. Taken together, the results highlight how contextual hostility not only shapes how refugees are treated, but also how they come to experience and navigate their place in society, reinforcing a broader process of social withdrawal and marginalization.
These findings carry broader implications for both theoretical and empirical research on refugee integration. First, conceptually, I introduce the concept of social marginalization into scholarly debates on integration and highlight its critical role in undermining psychological integration and, subsequently, broader social integration for refugee communities. Social marginalization represents a broader and often less studied dimension of refugee experiences, going beyond economic or civic integration to demonstrate how exclusion shapes everyday interactions. Theoretically, I draw on insights from social psychology to build an argument about how anti-refugee violence contributes to increased social marginalization. Empirically, this is the first study to analyse the effect of anti-refugee violence on social marginalization. While the link between violence and social marginalization might appear straightforward, this relationship has rarely been conceptualized in a systematic framework. By theorizing social marginalization as a distinct dimension of integration, the study demonstrates why even seemingly self-evident relationships require systematic theoretical development to be properly understood. In doing so, it contributes to the growing literature on how discrimination shapes the lived experiences of refugees in host societies (Fouka 2019; Abdelgadir and Fouka 2020; Theorin et al. 2021; Tyrberg 2021).
This paper is organized as follows. The first section presents the theoretical framework. I begin by reviewing the literature on anti-refugee discourse and its consequences for refugee communities, highlighting the gap in research regarding the relationship between exposure to anti-refugee violence and social marginalization. I then theorize how anti-refugee violence contributes to the marginalization of refugees. Following this, I introduce the case of Germany. The subsequent section outlines the data sources, followed by a description of the data merging strategies and the multilevel modeling approach used for estimation. I then present the main findings, before concluding with a discussion of their implications for broader debates on integration and anti-refugee attitudes.
Theoretical Framework
Despite extensive research on anti-migrant attitudes, the specific effects of anti-migrant violence on refugees and other migrant groups remain underexplored. Such violence often intersects with far-right ideologies. Consequently, an emerging body of scholarship has examined the ramifications of far-right hate crimes and xenophobic violence specifically targeting migrants. Steinhardt (2018), analysing the xenophobic violence against immigrants in Germany in the early 1990s, found that exposure to macro-level violence reduced several dimensions of socio-economic integration and increased their intention to return to their native country. Similarly, Gould and Klor (2016), in the post-9/11 American context, also found that Muslims living in states with higher hate crimes exhibited lower levels of assimilation. They had higher chances of marrying within their own community, higher fertility rates, lower labor force participation, and lower English proficiency. In contrast, Jaschke et al. (2022) found that refugees assigned to German areas with more local hostility towards them assimilated more quickly towards the local culture. However, their measure of hostility was an expansive index that combined the vote share of far-right parties, historical patterns of violence against Jews, and attacks against mosques.
While these studies have advanced our understanding of how hostile environments shape integration trajectories, the psychological experiences of migrants, especially refugees, remain comparatively understudied. Previous research has found that exposure to hostility and violence can have severe consequences for targeted minority groups, including migrants and racial minorities, leading to significant psychological distress (Galovski et al. 2016; McLeod et al. 2020; Alang et al. 2022). Therefore, it is reasonable to assume that exposure to anti-refugee violence will also affect the psychological and mental state of refugees. Nevertheless, there is an overall lack of studies that have analysed the effects of anti-refugee violence on the psychological state of refugees. This paper aims to fill this gap by analysing the effects of residing in geographic areas with high anti-refugee violence on social marginalization among refugees.
In this paper, I utilize the term anti-refugee violence to denote a range of violent activities and acts of aggression against refugees, including demonstrations, assaults, arson, and other miscellaneous attacks such as hostile protests or Nazi graffiti used by sub national or non-state entities against refugees. It is characterized by elements of xenophobia, nationalism, and racism (Miller 2017). Anti-refugee violence is not just a collection of isolated incidents but a powerful symbol of exclusionary practices that reflect broader patterns of pervasive discrimination against refugees.
Such violence must be understood not only as a physical threat but also as part of the broader social environment into which refugees attempt to integrate. Members of non-dominant groups can only pursue integration successfully when the host society adopts an open and inclusive stance toward cultural diversity (Berry 2011). However, this ideal is often not met. When the majority group imposes specific forms of cultural adaptation, limits minority agency, or exhibits outright hostility and exclusion, integration is no longer a matter of individual willingness or capacity alone. In such contexts, structural and symbolic barriers undermine the very conditions necessary for integration. An additional dimension must therefore be considered. This paper introduces and develops on the concept of social marginalization to capture how individuals perceive and respond to their rejection from mainstream society moving beyond traditional integration frameworks to account for hostile or exclusionary social environments.
Social marginalization is a multifaceted concept that has received relatively limited theoretical development in migration research. While rooted in social psychology, it has often been conflated with or subsumed under adjacent constructs such as social exclusion, psychological integration, and disidentification. Some researchers have used the terms “social exclusion” and “social marginalisation” interchangeably, defining social marginalization as exclusion based on one’s ethnic background (Hiott et al. 2006; Issmer and Wagner 2015). Others focus more narrowly on individual self-reports of exclusion from institutions or opportunities (See Aydin et al. 2010). More recently, Bollwerk et al. (2022), introduced the concept of perceived social marginalization, emphasizing group-level perceptions as stronger predictors of negative social outcomes than individual experiences. While their framework provides valuable insights, it overlooks individual experiences that may shape perceptions of one’s own group. This limitation underscores the need for a more comprehensive understanding of social marginalization that integrates both individual and group-level dynamics.
In this paper I define social marginalization as the subjective and cognitive perception that one’s own group is unappreciated, treated as unimportant (Bollwerk et al. 2022), and unwelcome in the society (Hiott et al. 2006). It is a subjective phenomenon and may be influenced by objective individual experiences or can be anchored in group level perceptions. Additionally, it is also distinct from the concept of economic marginalization which largely pertains to the economy-based perspectives on marginalization (Andersson 2003). The advantage of this definition lies in its nuance, it captures the interplay between individual and group-level experiences, offering a holistic perspective on how marginalization is perceived and experienced. By encompassing both personal and collective dimensions, this framework allows for a richer analysis of the social dynamics underpinning marginalization.
This conceptualization also differs from psychological integration, which typically captures the degree to which individuals feel affectively connected to the host society and envision their long-term future within it (Harder et al. 2018). In contrast, social marginalization pertains less about internal orientation and more about external positioning. It centers not on how individuals feel about the host society, but on how they perceive the host society to feel about them. Rather than reflecting a lack of identification, it reflects perceived rejection the belief that one’s group is not respected, accepted, or fully recognized as part of the national community.
Similarly, it must be distinguished from disidentification. Disidentification describes a psychological strategy of distancing oneself from a national identity in response to rejection (Jasinskaja-Lahti et al. 2009). It reflects an active motivational shift. By contrast, social marginalization does not imply rejection of the national identity but captures a sense of being denied membership in it. While experiences of discrimination may trigger both disidentification and marginalization, the former is a response rooted in identity distancing, whereas the latter captures the felt absence of social inclusion and belonging.
Social marginalization, as conceptualized here, is thus a reactive and involuntary outcome, not a freely chosen withdrawal. It emerges when refugees face social and rhetorical closure when avenues for participation are blocked by hostility, exclusionary rhetoric, or violence. It offers a more grounded account of how perceived societal rejection undermines belonging. By integrating both personal and group-level dimensions, this framework contributes a nuanced lens through which to understand the effects of hostile contexts such as anti-refugee violence on refugees’ perceived place in society.
However, the question arises: How does exposure to anti-refugee violence influence the refugees’ social marginalization? Previous research suggests that there is a distinction between group and individual level perceptions of discrimination (Fuegen and Biernat 2000; Oskooii 2016). Group level discrimination pertains to the discrimination against the group as a whole, while the individual level discrimination would be the discrimination that individuals experience on a more personal level but because of ties to a certain group. Building on this distinction, I argue that anti-refugee violence contributes to both, it increases refugees’ personal experiences of discrimination while simultaneously shaping the perception that their group as a whole is unwelcome in society. I hypothesize that these effects operate through two interconnected mechanisms: (1) The reorientation of social ties away from the majority group and toward co-ethnics, and (2) the strengthening of ethnic identity as a protective response to societal rejection. These mechanisms operate within a broader socio-political climate in which violence and exclusionary rhetoric work together to create a hostile normative environment.
Anti-refugee violence rarely occurs in isolation. It is often embedded in a broader surge of anti-migrant rhetoric within the socio-political climate (Hopkins 2010; Piatkowska et al. 2020). Although this rhetoric is nominally directed at migrants in general, it frequently targets refugees in particular. These two forces—rhetoric and violence—are mutually reinforcing, making it difficult to disentangle their individual effects on refugees. Violence signals to refugees that they are not accepted in the host society, while also legitimizing and amplifying exclusionary views in the broader public. This aligns with broader research showing that symbolic exclusion and negative majority attitudes even when not tied to direct, localized violence can undermine minority well-being and perceptions of belonging (Habtegiorgis et al. 2014). This shifting normative environment helps normalize exclusionary rhetoric, lending legitimacy to views that were previously confined to the political fringes. As a result, individuals who already harbor such beliefs may feel emboldened to express them more openly, manifesting as increased xenophobia and hostility in everyday life (Bursztyn et al. 2017; Bischof and Wagner 2019). In this way, anti-refugee violence is not just an expression of hostility; it functions as a public signal of group-level rejection and contributes to a broader hostile climate. Such a climate is not only symbolic but has tangible effects on how refugees experience both their group’s standing in society and their day-to-day interactions.
This normalization of exclusionary rhetoric at the macro and micro level can heighten interpersonal discrimination and perceived group level rejection experienced by migrants and refugees alike. Empirical evidence supports this link: For example, municipalities led by far-right mayors in Italy show significantly higher rates of hate crimes (Romarri 2020), Similarly, in the US context, Trump’s tweets about Muslims and Islam-related topics have been linked to subsequent increases in anti-Muslim hate crimes by his followers (Müller and Schwarz 2023). Together, these findings illustrate how rhetorical hostility and political legitimation can spill over into real-world violence and discrimination. Taken together, these dynamics suggest that anti-refugee violence, operating within a broader hostile climate, serves both as a symbol and a channel of exclusion, reinforcing group-level rejection and everyday experiences of discrimination.
Incidents of anti-refugee violence not only reflect a broader climate of hostility but also shape how refugees perceive and navigate their social environment. Research on discrimination distinguishes between isolated random acts and systemic patterns of group-based discrimination. Individuals respond differently when they perceive discrimination as rare and random versus when it appears persistent and targeted toward their group (Branscombe et al. 1999). Anti-refugee violence often spatially concentrated, well-organized, and frequent can appear to refugees not as isolated outbursts but as part of a broader pattern of pervasive exclusion. As a result, refugees may interpret these incidents as manifestations of underlying xenophobia and hostility within the host society, viewing them as a systemic and pervasive issue rather than isolated random incidents. Crucially, this interpretation of the hostility is important. Research shows that when individuals perceive group-based harm as intentional and illegitimate, they are more likely to respond with stronger emotional and behavioral responses (Rule and Ferguson 1984). When these acts are carried out by multiple members of the host society, refugees are more likely to generalize responsibility to the broader out-group, increasing their social distance from the native population. Such experiences communicate that the host society is not merely indifferent, but actively hostile, a message that shapes not only how refugees feel, but how they act in everyday social life.
The concept of stigma offers a valuable lens for understanding these dynamics. This climate of hostility, through both direct experiences and perceived group-level threats, acts as a cultural stressor (Cano et al. 2016), generating a sense of stigma for refugees. Stigma, in this context, is more than an individual mark; it is a structural and social process involving stereotyping, separation, status loss, and discrimination (Yang et al. 2024). These interconnected processes operate within broader structures of power, In this context, anti-refugee violence functions as a tool of that power, used to “keep refugees down, in, and away” (Pachankis and Wang 2024: 77). It symbolizes to refugees the hostility of the broader society and reinforces their sense of exclusion from the host community. The experience of stigma, rooted in societal rejection intensifies their feelings of marginality and directly harms their psychological well-being and social status (Link and Phelan 2001; Major and O’Brien 2005; Pachankis and Wang 2024).
Empirically, previous research confirms that discrimination and exclusion contribute to psychological distress and reduce trust in the broader society (Coker et al. 2009; Kauff et al. 2017). In contexts marked by frequent anti-refugee violence, refugees are likely to internalize a sense of vulnerability, which can erode their feelings of safety, belonging, and inclusion. Over time, this can foster a deeper sense of estrangement from the host society. This process of marginalization unfolds through two key mechanisms.
The first mechanism through which this unfolds is the reorientation of social ties: Refugees who perceive society as hostile may withdraw from contact with the majority population and increasingly rely on co-ethnic networks for safety and support. While this shift can offer short-term resilience, it may also reduce opportunities for intergroup contact, thus reinforcing marginalization. A second mechanism is the strengthening of ethnic identity itself. The rejection–identification model suggests that perceived discrimination strengthens in-group ties as a buffer against psychological harm (Branscombe et al. 1999). In-group identification thus becomes more than just a defensive posture it is also a restorative and relational response, allowing refugees to reclaim dignity and community in the face of social devaluation. While the theoretical framework emphasizes the broader, persistent differences in hostility across counties, the mechanisms through which such hostility operates such as changes in social ties and group identification may be most immediately visible in response to short-term spikes in local violence. Over time, these short-term reactions can accumulate, contributing to more durable patterns of marginalization observed across counties with consistently higher levels of anti-refugee hostility.
Taken together, these dynamics suggest that anti-refugee violence can shape both psychological and behavioral responses that cumulatively heighten refugees’ sense of social marginalization. I therefore hypothesize that refugees residing in counties with higher levels of anti-refugee violence will report greater social marginalization compared to those in less hostile areas (H1).
The German Case
To test the hypothesis, this paper focuses on the case of Germany, which meets two key scope conditions necessary for the argument: (1) A substantial refugee population, and (2) the presence of anti-refugee violence with subnational variation. Germany satisfies both conditions for several reasons. The Syrian civil war and the rise of ISIS in the Middle East led to the displacement of millions of people in Syria and Iraq. Germany and Sweden emerged as the two main destinations for these refugees in Europe. Angela Merkel, the German chancellor at the time, suspended the Dublin Regulation, an EU regulation that restricts the rights to asylum for refugees limited to the first EU member state they arrived in. This liberal refugee policy led to Germany providing temporary asylum to almost all the incoming refugees. Between 2015-2019 approximately 1.7 million people applied for asylum in Germany (Oltermann 2020). However, not all sections of the German society were equally welcoming to these policies. Political parties such as CSU that were part of Merkel’s ruling coalition objected to the policy. Similarly, Alternative for Deutschland (AfD), the far-right party in Germany, also openly challenged the policies and started increasing their vote share.
This political backlash was accompanied by a marked rise in violence targeting refugees, peaking in 2015 and 2016. The attacks ranged from xenophobic demonstrations and physical assaults to arson attacks on refugee shelters and symbolic acts such as painting swastikas on asylum housing. Importantly, these incidents were not confined to fringe regions but occurred across the country, with both urban centers (e.g. Berlin, Dresden, Cologne) and rural areas affected. Eastern Germany, in particular, saw a higher concentration of violent events, though significant incidents were recorded nationwide (Benček and Strasheim 2016). This broad geographic distribution and the range of hostile acts reflect a hostile integration environment that makes Germany an especially appropriate case for studying how contextual hostility shapes refugee marginalization.
Thus, Germany offers a context where both key conditions large-scale refugee arrival and significant subnational variation in anti-refugee violence are clearly met. In addition to meeting the scope conditions, Germany provides a methodologically advantageous setting for studying how local hostility shapes refugee outcomes. Crucially, refugees in Germany were subject to mobility restrictions that reduce concerns about self-selection bias, which often undermines observational studies. This restriction is analytically important. In contexts without such constraints, refugees could self-select out of high-violence areas, especially if those more sensitive to threat or violence (based on personality, demographics, or resources) had the means to relocate. Over time, this would result in more resilient refugees remaining in violent areas, potentially biasing estimates downward. Germany’s policies mitigate this concern by legally limiting mobility, allowing for stronger inferences about how exposure to violence affects the average refugee.
Refugees arriving in Germany were distributed across federal states using the Königstein Key a formula based on state tax revenue (two-thirds) and population share (one-third) (Geis-Thöne and Orth 2016). Newly arrived asylum seekers were first assigned to a state, then to a county, and finally to a municipality. In August 2016, a policy change stipulated that all refugees, including those with permanent residency or temporary permits, were not allowed to move from their initial state of residence for up to three years (Kanas and Kosyakova 2023). In 10 of the 16 federal states—Bavaria, Baden-Württemberg, Hesse, North Rhine-Westphalia, Saarland, Saxony, Saxony-Anhalt, Berlin, Bremen, and Hamburg, refugees were further restricted from changing counties for up to three years. Local politicians were thus unable to influence the inflow of refugees into their states and counties. Likewise, refugees faced legal limitations on their ability to choose the state or county in which they wished to reside. Consequently, they could not avoid areas with higher levels of anti-refugee violence.
Some studies argue that the allocation was not entirely random, noting that the availability of local accommodation may have shaped county-level placements (Gehrsitz and Ungerer 2022). While this may have introduced some practical constraints on randomness, Germany still offers one of the most suitable empirical contexts for this study. The analysis focuses on the 10 states where county-level residence restrictions were in place, reducing the likelihood of selection bias and supporting a robust correlational analysis of the association between anti-refugee violence and refugees’ social marginalization. While the design does not allow for strong causal claims, it permits a robust correlational analysis grounded in a plausibly exogenous distribution of refugees.
Research Design
The main independent variable in this study is anti-refugee violence, for which data is sourced from the ARVIG dataset. ARVIG is a geo-referenced event dataset documenting violence against refugees in Germany from 2014 to 2017. It includes precise latitude and longitude information for all recorded incidents, down to the municipal level. The dataset has been widely utilized in social science research (Frey 2020; Igarashi 2021; Eger and Olzak 2023; Kuhn and Maxwell 2024), underscoring its reliability and credibility for studies on anti-refugee violence. The dataset categorizes violent attacks into four types: Xenophobic demonstrations, arson attacks, physical assaults, and miscellaneous attacks on refugee housing, such as painting swastikas on shelters (Benček and Strasheim 2016). In total, the dataset records 4,886 incidents between January 1, 2016, and November 11, 2017. Incidents of anti-refugee violence were recorded across Germany, with certain areas experiencing a notably higher concentration of events. Larger cities such as Berlin, Dresden, and Cologne were focal points for such incidents, though rural regions were not exempt. While urban areas displayed a higher density of attacks, significant events also occurred in smaller municipalities. Overall, Eastern Germany exhibited a greater concentration of violent incidents compared to the western part of the country, although such events were observed nationwide.
I use data from the German SOEP, specifically the 2016–2017 waves, to measure the social marginalization of refugees. The SOEP is a nationally representative household panel survey conducted annually in Germany, with specialized samples for particular population groups. The M3 and M4 refugee samples are part of a joint project by the Institute for Employment Research (IAB) and the Research Centre of the Federal Office for Migration and Refugees (BAMF-FZ). These samples include individuals who arrived in Germany during the 2014–2017 refugee crisis and had applied for asylum by April–June 2016. The M3 sample includes individuals who arrived in 2015 and early 2016, first interviewed in 2016. The M4 sample consists of individuals who arrived in 2016, first interviewed in 2017. Respondents were recruited via the Central Register of Foreign Nationals (Ausländerzentralregister, AZR). Households were randomly selected from the register and contacted for face-to-face interviews conducted in multiple languages by trained interviewers. In the baseline wave (2016), more than 3,000 adult refugees were successfully interviewed. The response rate for the baseline refugee samples was approximately 50 percent, which is considered strong given the logistical and linguistic challenges of surveying recently arrived refugees.
The surveys were conducted annually from 2016 to 2019 and are designed to be nationally representative of asylum seekers in Germany. For the present analysis, I draw on the 2016 and 2017 waves, which provide two waves of repeated cross-sectional data on refugees’ social marginalization. The dependent variable is social marginalization. I employ a multifaceted operationalization approach and utilize three distinct but interrelated variables to capture social marginalization. The first variable, personal experiences of discrimination, is operationalized by the survey question, how often have you personally experienced being disadvantaged in Germany because of your origin (scale from 1 to 3)? This item reflects the individual-level dimension of social marginalization, linking everyday discriminatory experiences to broader perceptions of exclusion. The second variable, feeling like an outsider is operationalized by the survey question, how often do you feel like an outsider (Scale 1–5)? as previously also used by Hiott et al. (2006). The third variable, feeling welcome is operationalized by, do you feel welcome in Germany (Scale from 1 to 5)? Together, these two items capture symbolic and group-level perceptions of societal acceptance or rejection, reflecting how respondents perceive the host society’s stance toward their group as a whole, independent of specific discriminatory incidents. To facilitate the statistical analysis, all three variables have been rescaled to a unified continuous scale of 0 to 1 (the process is explained in the appendix).
In principle, the SOEP allows for panel analyses. However, the relevant items on social marginalization were only included every second wave, and data on anti-refugee violence are available only up to 2017. This mismatch makes a within-person design infeasible. The analysis therefore uses two cross-sectional waves, ensuring that respondents’ reported attitudes are temporally aligned with their contextual exposure to violence.
Although the mean levels of perceived exclusion and discrimination appear relatively low, this pattern likely reflects the composition of the sample rather than measurement artefacts. The dataset includes an oversample of recently arrived refugees from Syria and Iraq, most of whom had lived in Germany for a relatively short period at the time of the survey. Prior research shows that newly arrived migrants often report lower levels of perceived discrimination and higher levels of optimism compared to longer-settled or second-generation immigrants (Maxwell 2010; Röder and Mühlau 2011). Consistent with this interpretation, the descriptives nevertheless reveal meaningful variation: 21.15 percent of respondents report feeling like an outsider often or very often, 35.06 percent report having personally experienced discrimination sometimes, and around 16 percent indicate that they feel only partly, hardly at all, or not at all welcome in Germany. These figures demonstrate that, even among newly arrived refugees, a significant minority experience social distance and exclusion, underscoring the relevance of studying social marginalization in this population.
To examine potential behavioral mechanisms linking contextual hostility to social marginalization, I analyse changes in refugees’ social networks using two survey items. These items capture the frequency of interaction with both majority-group members and coethnics. Specifically, respondents were asked how often they (1) have contact with Germans in their circle of friends, and (2) spend time with people from their country of origin (not including family members). Response options range from Every day (1) to Never (6). For ease of interpretation, all three variables have been reverse-coded and rescaled to a unified continuous scale of 0–1 range, where higher values indicate more frequent contact. This allows for consistent comparisons across indicators and facilitates the estimation of linear effects.
Modeling Strategy
The theoretical focus of the paper is on contextual hostility. I argue that refugees living in counties with a higher typical level of hostility during the study period experience greater social marginalization than those in less hostile counties. To explore this cross-sectional hypothesis effectively, I use a modeling approach that separates the effects of long-term differences between counties from short-term changes within counties.
Operationalization of the independent variable
In line with Eger and Olzak (2023), I measure exposure to anti-refugee violence by calculating, for each respondent in the SOEP refugee sample, the cumulative number of violent events in their county during the 365 days leading up to their interview date. This rolling one-year window captures the local level of hostility close in time to when the refugee was surveyed, allowing me to compare respondents within the same survey wave who were interviewed at different points in the year. Because counties in Germany vary substantially in population size, I express this measure as incidents per 1,000 county residents. This population-weighting ensures that the variable reflects the relative intensity of violence rather than simply the raw number of incidents, making it comparable across both densely and sparsely populated counties.
Mean-centreing and decomposition
Following the logic of the theoretical framework, which distinguishes between long-term contextual hostility and short-term fluctuations, I apply group mean-centering to decompose this variable into two orthogonal components:
The between-county effect reflects the average differences in anti-refugee violence across counties over the observation period. The emphasis on between-county variation follows directly from the theory: Social environments differ across places, and these differences shape refugees’ experiences of belonging and exclusion. Put simply, the between-county effect tests whether refugees in more hostile counties report systematically higher levels of marginalization than those in counties with lower hostility. In contrast, the within-county effect captures short-term changes in hostility within the same county. It reflects whether refugees interviewed shortly after a local spike in anti-refugee violence report greater marginalization than those interviewed earlier or later in the same place. While the theory focuses on broader differences in hostility across counties (between-county effects), the within-county effect provides a supplementary test of whether changes in violence over time within a county also shape refugees’ experiences.
This approach, following Fairbrother (2014) and Schmidt-Catran and Fairbrother (2016) who emphasize correctly specifying the random effects structure to avoid biased standard errors allows me to estimate the distinct effects of long-term contextual hostility and short-term fluctuations in violence on refugees’ reported marginalization. I estimate four-level hierarchical models: Individuals nested in county-years, which are further nested in counties and then in German states. In line with Schmidt-Catran and Fairbrother’s recommendation, the models include random intercepts at the county-year, county, and state levels, while using fixed slopes. This structure accounts for unobserved heterogeneity at multiple levels and produces more accurate standard errors.
I include a set of individual-level and contextual-level controls to address potential confounding factors that may influence both the predictor (exposure to anti-refugee violence) and the outcome (social marginalization).
Individual-level controls: Gender, age, education, and employment status are standard sociodemographic controls in research on migrant well-being and integration, as these characteristics can affect both the likelihood of exposure to hostility (e.g., through visibility, occupation, or social networks) and refugees’ perceived levels of marginalization (e.g., through differences in social capital and integration opportunities). Partnership status and family separation are included because family and intimate ties can provide emotional and social support that may buffer against hostile environments. At the same time, these relational factors can shape exposure to hostility (e.g., mixed social networks) and influence perceived inclusion.
Contextual-level controls: County-level GDP and employment rate are included as proxies for local economic conditions, as economic prosperity and labor market dynamics can shape both anti-refugee sentiment and refugees’ opportunities for integration, thereby influencing their experiences of marginalization. The share of the foreign-born population is included because the local ethnic composition can influence the salience and incidence of anti-refugee violence (e.g., through visibility of minorities) as well as opportunities for intergroup contact, both of which may affect perceptions of social marginalization.
Full technical details, including the algebraic model and an illustrative example, are provided in the methodological appendix.
Results
Table 1 presents the results for the main independent variable anti-refugee violence, measured as the number of events per 1,000 residents in the past 365 days at the county level and its association with three dimensions of social marginalization: Feeling like an outsider, perceived discrimination, and feeling welcome. As hypothesized, refugees living in counties with higher levels of anti-refugee violence are expected to report greater social marginalization.
Multilevel Models (365-day Effects) for Three Dependent Variables: Feeling Like an Outsider in Germany, Personal Experiences of Discrimination, Feeling Welcome in Germany.
Standard errors in parentheses.
Model 1 examines the outcome feeling like an outsider, a self-reported measure of social marginalization. The coefficient for the between-county effect is positive and statistically significant (
Predicted Values of Feeling like and Outsider by Between-county Violence (365 days).
Model 2 reports the results for personal experience of discrimination. Again, the between-county effect is statistically significant and positive (
Predicted Values of Personal Experience of Discrimination by Between-county Violence (365 days).
Model 3 examines the results for feeling welcome. The between-county coefficient is negative but it narrowly misses conventional thresholds for statistical significance (
Predicted Values of Feeling Welcome by Between-county Violence (365 days).
To further probe the temporal dynamics of exposure, I estimate supplementary models using a shorter level of violence in 180-days window prior to the interview as a variable. These models, presented in Appendix Tables 2.4, 3.4, 4.4, offer theoretical nuance: While the 365-day models capture the cumulative impact of persistent contextual hostility, The 180-day models are better suited to capture recent spikes in violence, which may quickly heighten perceived threat and lead refugees to feel more excluded or adjust their behavior in the short term. The results show that 180-day models show a significant between-county effect for personal experience of discrimination, and a marginally significant effect (
The Intraclass correlations (ICCs) for the outcome variables are available in Appendix Tables 2.2, 3.2, and 4.2. While the ICC values are relatively small (ranging between X and Y), such magnitudes are typical for attitudinal outcomes in hierarchical data (Hummler and Ziller 2025; Eger and Olzak 2023; Eger and Breznau 2017). Even modest clustering indicates non-trivial dependence between respondents within counties and therefore justifies the multilevel specification. What matters theoretically is not the absolute proportion of variance at the contextual level, but how much of that variance can be explained by meaningful county-level characteristics such as anti-refugee violence (Eger and Breznau 2017).
Given these results, it is also important to clarify the rationale for the modeling strategy adopted in the paper. While county fixed-effects models could in principle be applied, they would absorb all between-county variation, the very source of contextual differences that this study seeks to explain. A multilevel specification is therefore more appropriate, as it explicitly models both within- and between-county components, aligning the analytical approach with the study’s focus on cross-county variation in contextual hostility. As a robustness check, models with state fixed effects were also estimated, and the results (Appendix Tables 2.11, 3.11, and 4.12) are substantively consistent with the main findings.
In addition to the core indicators of marginalization, I also examine whether anti-refugee violence shapes refugees’ social networks, a key mechanism through which marginalization may occur. Specifically, I assess two outcomes: (1) Friendships with Germans (model 4), and (2) contact with individuals from one’s country of origin (model 5). The main results, based on the 365-day specification, are presented in Table 2. These models reveal no statistically significant between-county effects for any of the outcomes. Among the within-county effects, contact with one’s own community reaches significance: Refugees interviewed during periods of heightened local violence in the preceding year report more frequent interaction with coethnic peers (model 5). This finding aligns with the interpretation that persistent contextual hostility fosters a gradual reorientation toward coethnic networks, even if other dimensions of intergroup friendship remain unchanged over longer periods.
Multilevel Models (365-day Effects) for two Dependent Variables: Friendship with Germans (model 4) and Contact with Own Community (model 5).
Standard errors in parentheses.
To further explore the temporal dynamics, supplementary models using a level of anti-refugee violence in the 180-days before the interview are included in the appendix (Tables 4.1.2, 4.2.2). These shorter-window models also offer theoretical nuance: They capture short-term volatility and provide additional insight into how hostility may trigger more immediate social responses. In these models, the within-county effects are statistically significant for all the outcomes. Refugees interviewed shortly after local spikes in violence report fewer friendships with Germans, alongside greater engagement with coethnic peers. These patterns suggest that reorientation of social ties may occur more acutely in the immediate aftermath of hostile events, even if only one of these shifts persists over longer time frames in the 365-day models.
Taken together, these findings provide tentative support for the proposed mechanisms. While the 365-day results remain the core of the analysis, the appendix models illustrate how short-term surges in violence may activate the very processes withdrawal from majority networks and greater in-group reliance that underpin the broader marginalization effects documented in the main text. These supplementary patterns are consistent with the theoretical expectation that marginalization emerges not from isolated events alone, but from the cumulative effect of repeated, short-term reactions to contextual hostility.
Conclusion and Discussion
In this paper, I examined the impact of anti-refugee violence on the social marginalization of refugees. The findings suggest that higher levels of anti-refugee violence are associated with increased social marginalization of refugees, as reflected in heightened personal experiences of discrimination and a stronger sense of being an outsider within German society. This article contributes theoretically to the literature on the impacts of anti-refugee violence by drawing insights from social psychology. In this paper I have argued that anti-refugee violence serves as a powerful symbol of the broader anti-refugee environment in the county. Violence is a direct and unambiguous way of communicating that the refugees are unwelcome in German society. Consequently, refugees living in areas with heightened violence are likely to be acutely aware of the pervasive anti-refugee climate. This influences their interactions with the host society and reinforces their experiences of social marginalization.
These findings contribute to the literature on migrant’s integration by introducing social marginalization as a distinct yet complementary lens with a particular focus on refugees. Existing studies have primarily focused on economic and civic dimensions of integration such as employment, education, or political participation or, more recently, psychological integration, which captures emotional attachment, belonging, and perceived acceptance (See, e.g., Gould and Klor 2016; Jaschke et al. 2022; Steinhardt 2018). Social marginalization, as conceptualized here, is not merely a lack of identification with the host society, as in disidentification, nor is it reducible to individual experiences of exclusion. Rather, it reflects how broader social climates marked by anti-refugee violence shape both individual and group-level perceptions of marginalization. By operationalizing this construct empirically and demonstrating its sensitivity to local violence, this paper extends the integration literature beyond behavioral and identity-based frameworks. It shows that when refugees perceive their group as unwelcome and rejected, this has concrete implications for how they experience inclusion, navigate social relations, and engage with the host society. In doing so, it shows how the concept of social marginalization can enrich our understanding of how exclusionary contexts undermine not just integration outcomes but the very sense of social membership.
Scholarship on the integration of immigrants and refugees has largely focused on individual level variables such as religion and demography (Heath and Martin 2013; Kogan et al. 2020; Reeskens and Wright 2014; Reitz et al. 2009) or government policies (Engdahl et al. 2020; Hainmueller et al. 2017) while the socio-political environment of the host country is still largely ignored. However, if refugees and other immigrants are attempting to integrate into a society with strong anti-immigrant attitudes, it is only natural that they respond by becoming more inward-looking and reducing contact with the natives. My results highlight this dynamic specifically in the case of refugees. These findings lend further evidence to the growing argument that members of a group will resign from public life in light of growing discrimination (Hobbs and Lajevardi 2019; Oskooii 2016). Given that the integration of refugees, specifically the integration of Muslims in wider society, is frequently questioned by policymakers and politicians, these findings underscore the need for creating analytical frameworks that analyse immigrant and refugee behavior and the socio-political climate of the host country together. It subsequently also brings into focus the growing academic debate surrounding the notion that successful acculturation and integration requires more acceptance from the host society (Deole 2019; Lajevardi 2020; Tyrberg 2021).
These findings also have implications for scholarship on discrimination. Previous studies on discrimination differentiated between perceptions of discrimination at the individual and group level. The former targets the individual based on their group affiliation, and the latter targets the group that the individual belongs to. These findings, however, bring to light a new argument: Increased group-level hostility can have a trickle-down effect and also increase everyday personal discrimination towards individuals of the outgroup. Increased anti-refugee violence in an area emboldens the locals who share these anti-refugee views. They begin expressing their views more openly and engage in more anti-refugee activities. More recent literature on hate crimes has also found that support at the macro level, in the form of a political party or a mayor, increased hate crimes against refugees and other minority groups (Bursztyn et al. 2017; Müller and Schwarz 2023), suggesting that in certain contexts and for certain groups, the distinction between group-level and individual-level discrimination is more interconnected.
These findings also underscore the limits of formal integration efforts when they are undercut by localized hostility and violence. Since 2015, the German state has invested over 40 billion Euros in refugee integration, yet such financial investments and legal inclusion frameworks may prove insufficient if daily life for refugees remains shaped by fear, threat, and social exclusion. The rise of the AfD as a major political force risks exacerbating this disconnect, as its growing influence may legitimize xenophobic attitudes and embolden acts of hostility against refugees. In such a context, refugees may receive the right to stay without ever feeling socially accepted or safe. The findings in this study suggest that reducing anti-refugee violence whether through community outreach, stronger law enforcement, or targeted prevention should be seen not only as a matter of public safety, but as a crucial part of effective integration policy. Although the focus here is on Germany, the underlying logic extends to other liberal democracies where rising far-right movements threaten to hollow out the social conditions necessary for meaningful refugee inclusion.
While this study offers new insights into how anti-refugee violence exacerbates refugees’ social marginalization, it also has limitations. I have used single point surveys to conduct the analysis. This limits the long term understanding of the effects of violence on refugees’ integration trajectory. However, future studies that use panel data could potentially shed light on more aspects, such as whether these effects are long term or they go away relatively quickly. Further, how does this social marginalization affect the other areas of lives for refugees. In particular, its effect on social and labor market integration and social trust would be particularly interesting to analyse. In the absence of adequate data this article remained limited in analysing these aspects. In addition, given that majority of the study’s respondents identified as Muslim, the findings underscore the need to examine how anti-refugee hostility intersects with broader dynamics of Islamophobia and racialization. These patterns are not limited to Germany, across Europe, visible minority groups often face stronger marginalization in areas where far-right movements are gaining support. Future work should examine how religion, visibility, and local hostility together shape the everyday experiences and integration outcomes of these communities.
Supplemental Material
sj-pdf-1-mrx-10.1177_01979183251414101 - Supplemental material for Anti-refugee Violence and Social Marginalization of Refugees: Evidence from Germany
Supplemental material, sj-pdf-1-mrx-10.1177_01979183251414101 for Anti-refugee Violence and Social Marginalization of Refugees: Evidence from Germany by Naman Rawat in International Migration Review
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental materials for this article are available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
