Fragmented along Substantive Dimensions

Social exclusion happens along several dimensions such that people can find themselves excluded in the realms of education, employment, housing, health, politics, and more – and often all at the same time. The field contains work that typically focuses on one substantive dimension of exclusion, but is fragmented across those dimensions, such that they tend to be considered in isolation. The resulting insights are valuable but incomplete. When individuals or communities are excluded along several dimensions, the effects of exclusion in one dimension feed into other dimensions, mutually reinforcing one another and amplifying the harms. Research that examines only one area does not adequately capture the cumulative dynamic of multiple intersecting planes of exclusion. Before I comment more extensively on this fragmentation, I will briefly present some recent work on hierarchies of exclusion from the disparate areas.

Scholars identify a number of exclusionary practices in education. In Belgium, Van Praag et al. (Reference Van Praag, Stevens and Van Houtte2016) observe teachers as they engage in assimilationist practices, variously discouraging ethnic minority students from speaking languages other than Dutch, instructing them to conceal their identities in order to succeed in life after they graduate, and safeguarding public spaces to ensure that they remain “culturally neutral” (1364). This purported neutrality is neither neutral nor inclusive, but decidedly Belgian – i.e., Turkish music is not acceptable, but Belgian music is – with the dominant aim of guaranteeing that students of Belgian descent do not feel excluded. In Georgia, work on language reforms shows that efforts to enable ethnic minorities to learn Georgian have reproduced a hierarchy of languages and with it an ideology of exclusion; nevertheless Georgian language acquisition among minorities has increased independently (Wigglesworth-Baker Reference Wigglesworth-Baker2018). Complementing this research, Berglund (Reference Berglund2016) shows that Georgian adolescents favor Georgian-speaking minorities over minorities that do not speak Georgian, but also that they favor Georgian-speaking Armenians and Azerbaijanis as much as or more than native Georgians, perhaps reflecting the above-mentioned hierarchy of languages.

The idea of ethnic hierarchies in exclusion has recently been explored in other realms as well, with research demonstrating that some groups are consistently more excluded than others. Zschirnt and Ruedin (Reference Zschirnt and Ruedin2016) conduct a meta-analysis of correspondence studies on ethnic discrimination in hiring and find that in European studies, Arabs and people of Middle Eastern origins face highest levels of exclusion, and Turks the lowest. People of Chinese, Indian, Pakistani, and Bangladeshi origins are in the middle, in that order. While ethnic hierarchies are typical in studies that include a number of different groups, they are also specific to place and time, such that in Austria, it is Austrian citizens of Nigerian ancestry who are most excluded (Weichselbaumer Reference Weichselbaumer2017). And among people born in the United Kingdom, non-White ethnic minorities – Pakistani and Bangladeshi men especially – fare worse than their White counterparts in obtaining high-quality jobs, which pay well and are secure (Zwysen and Demireva Reference Zwysen and Demireva2020).

Hierarchies of exclusion are further visible in research that examines shared and rental housing markets, which have been understudied. In the Greater London Area, discrimination against ethnic minorities in the shared housing market is common, but uneven; applicants with Eastern European names are the least disadvantaged, followed by applicants with Indian and African names, and then by applicants with Muslim or Arabic names (Carlsson and Eriksson Reference Carlsson and Eriksson2015). Discrimination levels are higher in areas of London with a high proportion of White British residents. In Ireland, applicants with Irish names are more likely to be invited to view an apartment than applicants with Polish names, who are more likely to be invited than applicants with Nigerian names; this hierarchy is exacerbated by gender such that Nigerian men are the most disadvantaged (Gusciute et al. Reference Gusciute, Mühlau and Layte2020). And, Van Der Bracht et al. (Reference Van der Bracht, Coenen and Van de Putte2015) find that Belgian landlords consistently discriminate against potential tenants who are Belgian citizens and speak Dutch without an accent, but who have Arabic-sounding names.

Finally, research on political attitudes about redistribution in Great Britain also reveals the majority’s hierarchical preferences regarding exclusion from welfare. Ford and Kootstra (Reference Ford and Kootstra2017) find that White Britons support income redistribution policies when they are presented as class-based, but not when they are framed in ethnic or racial terms. While White Britons are substantially less likely to support redistribution helping Black Britons, they are even less likely to support assisting Muslim Britons. Ford and Kootstra (Reference Ford and Kootstra2017) capture this drop in support both among White Britons who hold traditionally left-leaning political values that typically favor redistribution and among White Britons who do not express egalitarian political values.

Social exclusion does not stay confined to any one dimension. Some people may indeed experience exclusion only in employment, but many experience it in employment – and in education, and in housing, and in politics. What happens in one sphere rarely remains separate from the others; instead, these effects feed into and reinforce one another, alter individual behaviors, and have repercussions for the future. Thinking about human lives, it is often the same individuals who are most likely to be excluded from educational opportunities, the best jobs, and the best apartments, all the while not having sufficient political representation to achieve meaningful change. Importantly, the correspondence studies mentioned above demonstrate that exclusion happens when job or housing applicants are identical in all aspects except their ethnic identity. Yet by the time someone applies for housing, they may already have received inferior education and been passed over for a high paying job, which makes their candidacy less attractive to a potential landlord, even one that is not prejudiced. When we fail to consider the bigger picture, we miss the cumulative effects that exclusion produces – and this is where exclusion is most damaging.

The field therefore needs more work that examines what happens to people when the composite parts of exclusion work in tandem. Abdelgadir and Fouka (Reference Abdelgadir and Fouka2020) begin to bridge the dimensions of education and employment when they offer a causal assessment of the effects of the 2004 French headscarf ban on French girls who are Muslim. They show that exposure to the ban significantly reduced the girls’ likelihood of completing any secondary education and, over the long run, also negatively affected their labor force participation. They also show that the effects are expansive: the disruption to education happened via increased perceptions of discrimination in school, not merely among French Muslim girls who veiled prior to the ban but also among those who did not. More work could follow suit.

Fragmented on the Basis of Institutions and Behaviors

In addition to being fragmented along substantive dimensions, the literature is fragmented on the basis of what entities and actors contribute or react to exclusion. Scholars typically tackle institutions and individual behaviors separately. In real life, however, institutional exclusion does not happen in a vacuum; instead, it is accompanied by behaviors of both the majority and the minority, which have the potential to create dynamics that exacerbate exclusion or, potentially, break it. In this section, I briefly discuss work on institutions – specifically, the census – and then some work that examines individual behaviors. I finally turn to work that aims to bridge this divide by discussing two or more actors and entities at once.

Recent work on some of the fraught issues that Romani communities face around the census offers valuable insight into questions of visibility and inclusion in the context of institutions. Csata et al. (Reference Csata, Hlatky and Liu2020) use the case of Roma in Romania to show that municipal expert assessments are a sound validation strategy for censuses that rely on self-identification. Hay et al. (Reference Hay, Campbell, Kowalewska, Clark, Tammi and Balogh2020) detail the consequences of omitting Roma as an ethnic category from past census data collection in Scotland. They show that excluding Roma from data gathering processes leads not only to failures in identifying community needs and ineffective resource allocation, but also to their loss of participatory citizenship. They situate this invisibility alongside the hypervisibility that Roma communities experience as they are targeted by typically discriminatory portrayals by the media. While Hay and her coauthors recognize that collecting information about ethnicity is “complex and messy” (56), Surdu (Reference Surdu2019) digs deeper into the technologies that produce and reproduce the category of “Roma” for the purpose of census data collection. Of particular note is Surdu’s discussion on systemic surveillance practices by the police, which historically sought to categorize and count Roma and whose results were sometimes deemed superior to those of census agents even into the twenty-first century. Together, this work provides insight into the damaging and common coexistence of invisibility and hypervisibility, with theoretical and empirical implications that extend beyond Roma communities.

Terrasse (Reference Terrasse2019) also examines hypervisibility fueled by police surveillance, but instead in the context of individuals. She explores how police identity checks in France shape how French citizens who are second or later-generation immigrants view their own identities. Terrasse (Reference Terrasse2019) finds that respondents whom the police stopped two or more times within a single year are significantly less likely to identify as French and significantly more likely to identify with their minority identities. Also exploring identity strength, Van Heelsum and Koomen (Reference Van Heelsum and Koomen2016) demonstrate the resilience of second-generation Europeans of Moroccan descent to negative public discourse about them. Such discourse, in the public or the media, does not appear to weaken their national identity as Dutch, German, Swiss, Belgian or French; their ethnic identity as Moroccan also remains strong. Finally, exploring identity strength and loneliness, Madsen et al. (Reference Madsen, Katrine, Jervelund, Christensen, Gonneke, Walsh, Koushede, Nielsen, Due and Holstein2016) find that Danes who are second-generation immigrants are particularly resilient from loneliness when they identify with their ethnic group.

Scholars who study individual attitudes and behaviors, like those just mentioned, often focus on either the group that excludes or the group that is excluded. Some recent work, however, aims to capture both. One such set of studies does so in the context of European students who are second-generation immigrants. Jaffe-Walter’s (Reference Jaffe-Walter2016) work on the treatment of Danish students who are Muslim examines the dynamic between teachers who cloak assimilation pressures in expressions of concern and students who react to them. She aptly names this disconcerting combination of liberalism and assimilationist practice “coercive concern” (2). She finds that often teachers are not openly racist, but instead use the language of benevolent concern to aggressively push Muslim students to assimilate. Examples include teachers asking high school students about their sex lives, a humiliating experience for any high schooler, out of a compulsion to see Muslim students adopt liberal sexual values and behaviors more in line with Danish mainstream culture. Exploring student responses to this treatment, Jaffe-Walter documents retreat – students stop asking for help with coursework, avoid going on overnight school trips, and “distance themselves from Danish identities” (141). In Belgium, Van Praag et al. (Reference Van Praag, Stevens and Van Houtte2016) observe similar dynamics as teachers engage in “the cultural maintenance of ethnic minority students” (1364) by promoting activities seen as “culturally neutral,” meaning mainstream Belgian. In response, some students comply with teachers in order to protect themselves; others resist by speaking “Turkish in front of [the] teachers while leaving school grounds” (Van Praag et al. Reference Van Praag, Stevens and Van Houtte2016, 1364).

While most work focuses on one set of actors, studying multiple actors in tandem enables us to uncover the dynamic nature of exclusion: it is not merely individual actions or policies that matter in breaking or entrenching exclusion, but also their interactions, which can amplify the cumulative effects. Work by Bracic (Reference Bracic2020) takes a step in this direction. Bracic (Reference Bracic2020) develops the concept of an exclusion cycle in which anti-Roma culture gives rise to discrimination by non-Roma. Roma develop survival strategies to thrive in the space that discrimination has constrained. Non-Roma then resent these strategies, ethnicize them, and erroneously attribute them to Roma as such and not to their own discrimination, perpetuating anti-Roma culture and feeding the vicious cycle. This cyclical dynamic can be applied to other contexts such as that in Denmark explored by Jaffe-Walter (Reference Jaffe-Walter2016), where teachers pressure Muslim students to assimilate, the students react by retreating from situations where those pressures are particularly acute, and the teachers then erroneously attribute retreat to stereotyped Muslim cultural “traditions” (98) like insularity. Actions like discrimination or retreat matter in their own right, but the cumulative effect of their interaction is meaningful, too, as its self-perpetuating dynamic cements the rift between the two communities.

While this work has begun to address the dynamic interaction between majority and minority behaviors, a more complete picture would consider the institutional dimension as well. Garner (Reference Garner2019) offers such insight, exploring the interplay between state institutions, the media, and minority behaviors in the context of racialization of Gypsy-TravellersFootnote ¹ in England (admittedly, behaviors of the majority are less explored). Planning regulations, an institution that governs the use of space in England, were previously neutrally framed but nevertheless produced outcomes that disadvantaged non-sedentary communities. Recent changes to the law, however, explicitly single out Gypsy-Travellers and thus not only result in deliberate differential treatment but also seek to determine who is a Gypsy-Traveller. Garner studies this institutional mechanism, bolstered by the media that fuel anti-minority sentiment, alongside the strategies of Gypsy-Travellers. Some Gypsy-Travellers respond to mobility restrictions by moving into settled housing; others resist by purchasing land and using it as a “communal halting site, effectively sub-dividing the land, then applying for retrospective permission” (524).

Fragmented on the Basis of Methods

In addition to its fragmentation along the substantive dimensions of exclusion and on the basis of institutions and behaviors that contribute to it, the field is also methodologically fragmented, with most work using either quantitative or qualitative methods but not both. This fragmentation is problematic not least because it typically corresponds to substance. Work that focuses on elements that exclude, be they institutions or individuals, tends to be predominantly quantitative, while work that focuses on the excluded tends to be predominantly qualitative. To the extent that scholars engage more deeply with work that uses their preferred methods, the overlap between the methodological and the substantive split carries implications for how scholars ask questions as well as for the theoretical insight they rely on when forming arguments.

Work on hiring, for example, tends to use quantitative methods. The correspondence studies already featured in the preceding sections, like the study by Zschirnt and Ruedin (Reference Zschirnt and Ruedin2016) showing an ethnic hierarchy in which employers place candidates of Arab or Middle Eastern origins on the bottom, all offer examples where exclusionary decision-making is captured using quantitative approaches. In addition, Li and Heath (Reference Li and Heath2020) explore longitudinal effects of unemployment in the United Kingdom, showing that members of ethnic minorities, once unemployed, are more likely to suffer cumulative, long-term disadvantages both in terms of reemployment and pay. Moving beyond unemployment, Falcke et al. (Reference Falcke, Meng and Nollen2020) use cross-sectional data collected over a number of years to show that compared to Dutch graduates of non-immigrant descent, second-generation immigrants in the Netherlands with an applied science university degree are more likely to be excluded from jobs that match the type of their education, and, in the case of those who are of non-western minority descent, also the level of their education.

A subset of quantitative studies on exclusion use experimental methods. Exploring discrimination in retail, Bourabain and Verhaeghe (Reference Bourabain and Verhaeghe2019) conduct a field experiment in Belgium capturing whether salesclerks in stores greet, help, and surveil customers. Salesclerks greeted and helped customers of only Belgian descent significantly more often, but surveilled them substantially less than they surveilled customers who had at least one grandparent born in Morocco, Algeria, or Tunisia. Minority men were surveilled most. These findings speak to the twin phenomena of invisibility and hypervisibility mentioned above, showing that this pattern manifests itself in everyday contexts beyond policing. In the context of Roma exclusion, Bracic (Reference Bracic2016) employs a lab-in-field experiment in Slovenia and Croatia to capture discriminatory behaviors by non-Roma and so explores the effectiveness of two mechanisms geared at improving Roma inclusion. She finds that the EU accession process – an otherwise powerful human rights improvement mechanism – does not appear to reduce discrimination, but that Roma-led NGO action that promotes intergroup contact and dialogue appears to help. Simonovits et al. (Reference Simonovits, Kézdi and Kardos2018) experimentally test an alternative intervention that aims to reduce anti-Roma sentiment in Hungary. They find that participating in a perspective-taking online game reduces prejudice against Roma; the effects persist after a month, extend to attitudes towards refugees, and decrease support for an overtly racist party, Jobbik.

While quantitative approaches predominantly appear in work that examines elements that exclude, qualitative approaches are more common in research that focuses on excluded communities. A set of qualitative studies using ethnographic and interview-based methods richly characterize minority strategies and behaviors related to exclusion. Observing Roma, Traveller, and Gypsy students in the United Kingdom respond to exclusion in education, Ryder (Reference Ryder2017) divides them into the resisters, the semi-accommodated, the mainstreamers, and the assimilated. The semi-accommodated group of students strategically maintain a balance between occasionally engaging in resistance behaviors while generally keeping in with the “school ethos” (46); the mainstreamers instead exhibit high school conformity and suffer a narrowing of ingroup social networks. The assimilated do not identify as Roma, Traveller, or Gypsy, while the resisters use strategies of distancing and withdrawal. Ellefsen and Sandberg (Reference Ellefsen and Sandberg2021) conduct in-depth interviews with 90 young Muslims in Norway and uncover actions of everyday resistance: talking back, talking down, entering dialogue, living the example, and denying significance. And, Verwiebe et al. (Reference Verwiebe, Seewann, Wolf and Hacioglu2016) interview Austrians of various Balkan ethnicities to show that they react to exclusion in employment by selectively overperforming, avoiding situations where they might encounter discrimination, ignoring prejudice by normalizing it, using irony to deescalate, or resisting.

Assimilation and concealment are strategies that aim to erase differences, to varying degrees of success. Relying on a year of ethnographic fieldwork, Golubović (Reference Golubović2020) describes how Serbian women in Sarajevo practice concealment by making small changes to their body language in order to avoid being perceived as Serb. Assimilation is a strategy available to some, but not all, as demonstrated by Rodríguez-García et al. (Reference Rodríguez-García, Solana, Ortiz and Ballestín2021) in the context of Spain where biracial Spanish-born youths who more closely resemble the majority based on phenotype, language, or religious affiliation can develop identities that are not excluded, while Spaniards whose heritage is visibly different cannot. The ethnic Laz in Turkey likewise engage in assimilation, but also practice non-threatening boundary making in informal settings (Serdar Reference Serdar2019). Both Rodríguez-García et al. (Reference Rodríguez-García, Solana, Ortiz and Ballestín2021) and Serdar (Reference Serdar2019) collected their data using semi-structured interviews.

A number of works discuss collective strategies that mobilize. Initially on the margins of both women’s and Roma movements, Roma women in Bulgaria and Romania have developed their own, distinct intersectional activism; using semi-structured interviews, D’Agostino (Reference D’Agostino2021) offers a comparison, by state, of their mobilization on a number of dimensions. Stenroos (Reference Stenroos2018), an ethnographer, discusses the activism of Finnish Kaale Roma in the context of national Roma policy practices in Finland, outlining two systems of Roma power – the first stems from secular and traditional Roma customs and the second from the Kaale Pentecostal movement. Eijberts and Roggeband (Reference Eijberts and Roggeband2016) rely on in-depth interviews to show that the creation of safe spaces is an important collective strategy of women of Turkish and Moroccan descent in the Netherlands; in addition to being a welcoming environment, these spaces help women build and expand their social networks.

Finally, some recent work in the field bridges the methodological divide and employs both quantitative and qualitative approaches. For example, when Abdelgadir and Fouka (Reference Abdelgadir and Fouka2020) examine the educational and employment consequences of the French headscarf ban on French Muslim girls, they first use semi-structured interviews with some of the young women affected to develop their expectations, and then use survey and census microdata to test those expectations. Linos et al. (Reference Linos, Jakli and Carlson2021) use an experiment to show that Greek study participants donate only half as much to an NGO that serves free lunches in public schools when the donation appeal features a Roma child, compared to an ethnically Greek child. They complement their findings with interviews of elementary school principals. Bracic (Reference Bracic2020) uses a combination of lab-in-field experiments that include a videogame, ethnographic observations, and interviews with activists, public officials, and Roma and non-Roma citizens to explore the dynamic interaction between discrimination (non-Roma) and survival strategies (Roma). Lastly, Mares and Young (Reference Mares and Young2019) combine a wealth of information – several survey-based experiments, ethnographic observations, and over six hundred interviews with politicians, voters, and brokers – to explore the uses of different clientelistic strategies in rural contexts that vary in levels of economic precarity and share of Roma voters.

The studies just cited take up a small slice of the field; most work is either quantitative or qualitative. While some of the reasons behind this divide are quite legitimate, an inadvertent outcome may be that there is not as much engagement across the lines as there could and should be. Scholars may favor and thus engage more fully with the work methodologically closest to their own (as an experimentalist, I have certainly committed this error in the past). In this instance, such preferences may result in insufficient engagement with work that examines the other side, such that scholars who study mechanisms that exclude may engage less with the literature on the excluded, potentially missing insight that could be critically relevant to their own work (and vice versa).

The concept of assimilation is perhaps the clearest example of such a gap in understanding. Qualitative work described above details assimilation and concealment as strategies that members of some communities use in response to exclusion. This scholarship clearly identifies assimilation as a strategy of coping or survival, and not as a desired end. Yet some quantitative work misses this insight and thus does not interrogate how positioning assimilation as a positive goal may serve to replicate marginalization. Limited engagement with work from different methodological traditions has implications beyond mere awareness. Suppose a scholar is examining how an excluding majority behaves in response to a strategy of resistance by the minority. If scholarship exploring that particular strategy of minority resistance is methodologically distant, the scholar may miss its theoretical insight. Correspondingly, their research would be less likely to fully consider why the minority resists in a particular way – which is probably important to understanding how the majority misinterprets that behavior, and thus why the majority does what it does. Keeping up with theoretical advances across methodological divides is likely to lead to new insight faster; collaboration across the lines may hold even more potential for growth.