Discrimination can take place in all spaces and places where people interact. However, a careful look into the large body of empirical work that can be grouped under the heading “discrimination research” suggests that the concepts, theories, and methods employed vary significantly across studies. This variation is not simply a matter of the individual researchers’ likes and dislikes regarding concepts, theories, or methods. Both the forms of discrimination and how it can be measured vary across social domains, depending on whether the domain in question is based primarily on what we here coin “systems of differentiation” or “systems of equality”. Social domains that involve some kind of market transaction – such as employment or housing – are heavily dominated by processes of selection and differentiation. By contrast, social domains such as schools, health systems or public services should, in essence, provide all individuals with equal assistance. The different logics inherent in systems of differentiation and systems of equality have implications for the forms of discrimination located and the conclusions reached in studies.

This chapter builds on the distinction between systems of differentiation and systems of equality, reviewing a selection of studies of discrimination in various social domains. It does not aim at providing an exhaustive review of existing research, but to group studies according to the type of social domain in which discrimination occurs. This way of categorizing research demonstrates that there is an interesting interplay between social domains and their respective rationale (differentiation/equality), the types of methods employed and the forms of discrimination detected. The chapter concludes by a critical reflection on the ability of social science research to capture forms of discrimination that are less easy to spot.

5.1 Systems of Differentiation vs. Systems of Equality

Most market transactions involve some kind of differential treatment. When applying for a job or trying to rent an apartment, individuals normally compete with others in a more or less open market. To get access to the goods in question they need to appear qualified and attractive to employers or landlords – who can choose from a pool of candidates based on a set of formal and informal criteria. Sometimes, these criteria are quite formal and explicit – such as in many advertisements for vacant jobs – while in other cases, the criteria are informal and implicit, such as in the private housing market. Although both employers and property owners in most countries are bound by law not to discriminate against individuals based on characteristics such as race or ethnicity, market transactions of this kind nonetheless include selection, and thus an element of differential treatment, since not all applicants can rent a home or be offered a job. One or a few will always be granted access to particular goods at the expense of others who want the same. Whether or not this unavoidable differentiation is discriminatory depends on whether the choice of candidate is based on legitimate or illegitimate criteria; that is, whether the decision is based on formal qualifications or, say, influenced by the racial appearance or ethnic background of the candidates.

The element of differential treatment that is inherent in most market transactions does not exist in a similar manner in all social domains. The school system, for example, shall provide an education of good quality to all regardless of ethnic background or other characteristics. Likewise, public bodies such as health services or welfare offices shall offer equal services to increasingly diverse populations. Of course, direct discrimination may occur in these social domains as well. For example, teachers may favor students who share their ethnic background or religious beliefs and let this in-group favoritism come to the disadvantage of students of other ethnicities or religions. Similarly, welfare workers or public advisors might provide members of minority groups with less information about their rights to social benefits, for instance, because of a more or less conscious perception that certain groups are “less deserving” of public goods than others are. Yet the modus operandi in systems of equality is not selection. Individuals or groups do not compete over access to scarce goods similar to labor or housing markets. In fact, in systems of equality, market transactions (at least ideally) do not play a role at all. The absence of differential treatment as a key form of human action in systems of equality might suggest that direct discrimination is less prevalent. At a minimum, discrimination in such systems is less clear-cut than in systems of differentiation, and it is far more difficult to detect because the interaction takes place in spaces where researchers’ direct access to relevant processes of the interaction is limited.

This somewhat schematic distinction between systems of differentiation and systems of equality is useful when assessing the methods and theories used and the forms of discrimination most frequently reported in different strands of research. However, what goes missing in the distinction is social domains characterized by law enforcement, such as the police, customs, and the judiciary system – what could perhaps be coined “systems of control”. These are all social domains that rest on principles of equality for the law, yet extant research suggests that differentiation based on ethnic and racial appearance – what is often called “racial” or “ethnic profiling” – indeed takes place, for example in identity checks (e.g., FRA 2010; Jobard et al. 2012). In the remainder of this chapter, however, we will stick to the simple distinction between systems of differentiation and systems of equality, as the main intent is to show how the logic or functions of these two systems shape our knowledge about the prevalence and forms of discrimination.

The European Union Agency for Fundamental Rights (FRA) defines ‘discriminatory ethnic profiling’ as ‘the practice of basing law enforcement decisions solely or mainly on an individual’s race, ethnicity or religion’ (FRA 2010, 6). According to Robert Staples (2011), in the US context, the phenomenon in which racialized minorities are exposed to racial profiling dates back to the age of slavery and the awareness and critique of such practices have been present for decades. In Europe, the awareness of ethnic profiling is more recent, and often connected to policing and especially to counter-terrorism enforcement in the aftermath of 9/11, 2001, and later terror attacks in cities such as Madrid (2004), London (2005, 2017), and Paris (2015, 2016). Empirical studies have substantiated the biases in policing and sanctioning against ethno-racial minorities, such as in France, where an experimental survey in two main transportation hubs in Paris found that blacks were between 3.3 and 11.5 times more likely than whites to be stopped, and Arabs between 1.8 and 14.8 more times (Goris et al. 2009). This study shows that young men with a minority background who wear “urban style” cloths are targeted at particularly high rates.

5.2 Discrimination Research in Systems of Differentiation

In domains where gatekeepers regulate the access to certain goods based on competition between individuals– such as jobs in labor markets and rental contracts in housing markets – discrimination can be directly assessed by experimental approaches, and particularly by field experiments. The virtue of field experiments was explained in Chap. 4: By manipulating information about fictitious applicants’ race or ethnicity, while holding all other information constant, such studies allow the researcher to measure the direct effect of the chosen characteristic on the relative chance of being invited to a job interview, getting an offer of renting an apartment or getting a mortgage loan offer, compared to equally qualified native-majority applicants. Indeed, field experiments have proven very efficient in documenting the prevalence of discrimination in various social domains, yet almost exclusively in social domains characterized by systems of differentiation, such as labor, housing, and product markets, where the researcher can intervene in naturally occurring selection processes.

In a field experiment of housing discrimination in Italy, for example, Baldini and Federici (2011) investigated whether individuals of different gender and ethnic backgrounds are discriminated against when trying to access the rental market. The authors created twelve fictitious individuals – four with Italian-sounding names, four with Arab/Muslim names and four with East European-sounding names – and sent emails from these individuals to vacant rental apartments in altogether 41 Italian cities. In total, more than 3600 emails were sent in response to vacant apartments. The results show that, on average, Italian-named individuals received a positive reply from landlords in 62% of the cases, while the Arab- and Eastern European-named individuals received positive responses in 44% and 50% of the cases, respectively. These differences are all statistically significant. The results further show that discrimination is higher against male foreign names, in particular for the Arab-named group. Further, discrimination against foreign names appears to be higher in Northern Italy than in other parts of the country, again particularly against Arab males (for reviews of all field experiments of housing discrimination, see Auspurg et al. 2019; Riach and Rich 2002; Rich 2014).

Similar findings are recorded in field experiments in the labor market, where the researcher typically creates fictitious job applicants with ethnically distinguishable names. Bursell (2014), for example, sent more than 3600 job applications to vacant labor market positions in Sweden. The fictitious male and female job applicants had Swedish, Arabic, and North African names, but had identical qualifications. Bursell found that the foreign-named applicants had significantly lower chances of receiving callbacks for job interviews: The overall relative callback rate was 1.8, meaning that while the Swedish-named applicants had to apply for ten jobs before being contacted by an employer, the foreign-named applicants had to apply eighteen times to receive a callback. The study shows no differences in callback between Arabic-names and North African-named applicants, but for both minority groups, male applicants received far fewer callbacks than female applicants (for reviews of field experiments of employment discrimination, see Riach and Rich 2002; Rich 2014; Zschirnt and Ruedin 2016; Quillian et al. 2019).

While the vast majority of field experiments are conducted in the labor and housing markets, researchers have also used this method to study discrimination in other market places, such as in sales (Rich, 2014). A recent example is Bourabain and Verhaeghe’s (2019) study of discrimination against women and ethnic minority customers while shopping in clothing stores in Belgium. The authors conducted an in-person audit in more than 300 shops in which men and women with Belgian and Maghrebi descent asked salesclerks for help. The study shows that customers of Maghrebi descent received unfavorable treatment in comparison to their Belgian peers when asking for help, while also experiencing fewer greetings and more surveillance by salesclerks. Further, the study demonstrates that men are significantly more greeted and approached than women within both the Maghrebi and Belgian groups and that the intensity and form of discrimination tend to be subtler and lower in high-end than in low-end stores. This example shows that researchers are able to detect subtle forms of discrimination even in market transactions not characterized by selection. However, the fact that even sales interactions play out in more or less open market arenas makes even this social domain available for researchers’ experimental intervention.

All of these experimental studies have in common that they are conducted by researchers’ intervention in market interactions that were naturally taking place. Indeed, the open nature of market interactions is a precondition for such studies to be executed, since it allows the researcher to assess discrimination directly without running much risk of being “caught in the act” of deception (cf., Yinger 1986). Especially when investigating discrimination in labor and housing markets by using field experiments, researchers intervene in processes that are bound to be selective. This enables “clean” estimates of discrimination against specific target groups, everything else being equal.

Although field experiments have proved important in demonstrating the prevalence of discrimination in the access to employment, housing, and product markets, research in this tradition has however seldom engaged in the broader literature on ethnic and racial discrimination, including theories aiming to explain the formation, persistence, and reproduction of inequality. Indeed, field experiment research has usually dealt with only the first set of individual-level explanations presented in Chap. 3, typically revolving around the traditional distinction between taste-based and statistical discrimination, though at times also discussing the relevance of stereotypes, organizational cultures, and sociological notions of group positioning. One explanation of why structural-level theories are regularly absent in field experiment research might be that it is hard to assess how and when structures of inequality translate into actual selection decisions. Another explanation is simply that field experiments focus on the very first stage of a market interaction – the submission of a job or housing application – and not on later forms for interaction, such as day-to-day relationships between colleagues and managers at the workplace. Indeed, the very existence of selection processes in social domains where candidates compete in an open market allows the researcher to focus mainly on the extent to which discrimination occurs, rather than on why and how racial appearance or ethnic background come to matter in these very same processes.

5.3 Discrimination Research in Systems of Equality

Research on discrimination in social domains characterized by systems of equality, such as schools, health care, and public services, stands in contrast to the above-mentioned studies. Most importantly, research on discrimination in such domains are almost exclusively based on indirect measures, either by assessing ethnic inequalities at the aggregate level by the use of the residual method, or by studying more subtle acts of discrimination by the use of qualitative approaches, such as in-depth interviews and participatory observation. In both cases, the measure of discrimination is less clear-cut than the differential treatment of otherwise similar individuals found in experimental studies. However, the in-depth study of discrimination, which is especially found in qualitative approaches, has other important merits, such as the ability to analyze the findings in light of theoretical frameworks based on broader structures of inequality. Both in terms of methods and theories and the forms of discrimination detected, studies of discrimination in systems of equality consequently differ from studies of discrimination in systems of differentiation.

A typical example of this research tradition is Çelik’s (2015) study among male second-generation Turkish students in Germany. The students were participating in a vocational preparation program offered by the public labor office, and Çelik bases his study on a combination of semi-structured in-depth interviews and 6 months of participatory observation of everyday life in school. Although the interviewed individuals vary greatly in their general perceptions and opinions, they all had a strong sense of being part of a group that is systematically discriminated against, and they all had personal experiences of discrimination. The students shared a feeling that both teachers and school advisors treated them differently than majority German students and other minority students attending the same program, and that this differential treatment was due to stereotypes about young Muslim men of Turkish descent in Germany. According to Çelik, these experiences led the students to develop a reactive ethnic identity, constituted by a positive collective identity among themselves and an oppositional identity vis-à-vis majority society.

Another example is Farris and Jong’s (2014) large-scale study of second-generation young women of North African and South Asian descent in Denmark, France, Italy, the Netherlands, Spain, and the UK. The study aims at disentangling the various transitions from education to work and builds on both secondary analysis of national and regional statistics and on in-depth interviews with second-generation women, ethnic community representatives, non-governmental organization (NGO) representatives, teachers, and vocational/career advisors. Employing an intersectional framework of analysis, Farris and Jong show that although there seems to be a female advantage in the educational system, career advice offices and ethnic social networks tend to channel second-generation girls toward those jobs that are “reserved” for immigrant women, such as cleaning services and care work. The authors thus argue that research on discrimination needs to acknowledge the “discontinuity” of axes of inequality, suggesting that categorical membership such as gender, race, and class come to play differently in different contexts, institutional settings, and time periods.

A final example is Hedlund and Moe’s (2010) study of how indigenous people are met in the health care services in Norway. Building on in-depth interviews with Sámi women and men as well as with health and welfare professionals in rural areas where the Sámi represent a considerable minority, Hedlund and Moe demonstrate how the lack of cultural sensitivity and cultural competence among majority professionals in practice may lead to indirect discrimination of Sámi patients and clients. The Sámi in Norway, who for a long period of time were forced to assimilate into Norwegian society, maintain a strong historical memory and ties to the indigenous community. The authors argue that because these ties and memories are typically awoken in interactions with social health and welfare professionals who originates from the majority culture, the health and welfare services need to develop a cultural sensitivity to be able to provide for accommodated services and assistance for indigenous people. Interestingly, this study points to the distinction between direct and indirect discrimination, discussed in Chap. 2: In public services, treating different people as if they are similar may in practice be discriminatory. In the case of Norway, knowledge about the century-long history of structural domination of the Sámi minority is a precondition for providing adequate service and help to a population that lacks trust in the state apparatus.

These different studies show how individuals of various minority origins may experience subtle acts of discrimination in social domains characterized by systems of equality, such as schools and welfare services. The studies also demonstrate how minority individuals often interpret their experiences in light of broader structures of categorical inequality, such as being ascribed membership in Muslim or indigenous groups in Europe. Importantly, these subtle forms of discrimination detected by qualitative researchers are not readily accessible by other methods. Of course, differences in access to education or health services are detectable by statistical data, and studies using the residual method often provide strong indicators that differential treatment in systems of equality do occur (e.g., Babyar 2018; Heath and Brinbaum 2014). Qualitative studies of people’s experiences, however, are necessary to explore the role discrimination plays out in micro-level processes in schools, health care, and public services. Moreover, experimental data is generally lacking in these domains, mainly because it is difficult to conduct field experiments of discrimination where intervention in a selection process is not an option. The result is that research on discrimination that occurs as part of everyday interaction in schools or in encounters between minority individuals and workers within the health and social services differs quite fundamentally from research on discrimination in social domains characterized by systems of differentiation, methodologically, theoretically, and conceptually.

5.4 Implications

Research on discrimination in systems of differentiation tends to focus empirically on the extent to which discrimination occurs in selection processes, and theoretically on whether discrimination is caused by individuals’ racial animus or statistical uncertainty. Research in the system of equality, on the other hand, tends to focus on more subtle processes of stigmatization and exclusion, and it more often engages with structural-level theories of inequality. Although there exist many exceptions to this rule, in general, these two strands of research can be clearly distinguished in terms of both empirical focus and the theoretical perspectives employed. The questions are: Why is this the case – and does it matter?

The main explanation of why experimental approaches dominate research in systems of differentiation while seldom are used in systems of equality is that the modus operandi differs between domains. In domains characterized by systems of differentiation, selection processes regulate the access to goods and resources, and ultimately to power. In domains characterized by systems of equality, access to goods shall be provided to everyone who has a legitimate need for equal services. This basic distinction helps explaining why two distinct strands of discrimination research have developed, and why the dividing line between the strands not only goes between researchers’ preferred choice of methods but also between the social domains in question.

Importantly, the distinction between the two different system logics has consequences for the conclusions reached by research. In systems of equality, the absence of differentiating processes in which a pool of individuals compete for scarce goods means that researchers often cannot assess the direct role of discrimination by using field experiments. As field experiments are considered the gold standard in discrimination research, this implies that research cannot provide “clear and convincing evidence” (cf., Fix and Struyk 1993) of discrimination in systems of equality. By implication, conclusions drawn by research in systems of equality are deemed “uncertain” because – as shown in Chap. 4 – other methods suffer from limitations when the task is to investigate the prevalence of discrimination.

The reverse problem exists in systems of differentiation. Because researchers do have access to selection processes it is a relatively easy task to detect discrimination by conducting field experiments, thereby assessing the extent to which discrimination takes place. However, although selection regulates access to social domains such as the labor market and the housing market, these social domains – and especially the labor market – also consist of a range of everyday encounters, for example between colleagues at the workplace. Of course, discrimination may take place in these encounters too and there exists a large literature on workplace bias (e.g., Bielby 2008; Brief 2008; Wrench 2007). Yet because these interactions are not readily available for experimental intervention, research on subtle acts of discrimination in the workplace is far less prevalent than research on discrimination in the access to the labor market.

Because different methods provide different information about the type of discrimination that occurs, it is difficult to compare the extent of discrimination across social domains. This point brings us back to Reskin’s (2012) observation, namely that there is a lack of studies which investigate patterns of disadvantage across different areas of social life and how disadvantage may cumulate over time and space (see also Blank et al. 2004; and this book’s Chaps. 2 and 3). One reason why such studies are so rare is the fact that while discrimination is easily detected in social domains characterized by systems of differentiation, it is harder to uncover the discrimination that de facto occurs in social domains characterized by systems of equality.