This chapter examines the literature regarding the impact of racism on visible minorities. Racism, as defined by the European Commission against Racism and Intolerance, is at the origin not only of the overtly discriminatory actions of a few isolated individuals but also of more subtle, often unconscious biased behaviours that can cumulate and obstruct key life areas. The literature review suggests that bias in textbooks, discriminatory behaviour among educators, and peer bullying can hinder educational trajectories. Discrimination in accessing work-based learning opportunities during formal education, such as internships and apprenticeships, along with bias in law enforcement, may impede effective school-to-work transition. Pervasive discrimination in hiring and employment, as well as in the housing market, further exacerbates these challenges. Moreover, discrimination can negatively impact mental and physical health, possibly affecting individuals from in utero to old age, with bias among healthcare providers potentially worsening health disparities.
Monitoring and Assessing the Impact of National Action Plans Against Racism
Abstract
Academic analyses find that racism can impede the educational trajectories of visible minority students through multiple channels. Notably, these include underrepresentation in children’s books and secondary school textbooks, bias among educators, and bias among classmates.
Studies find that although there has been progress in addressing the quantitative and qualitative underrepresentation of visible minorities in children’s literature, there is evidence suggesting that it remains a prevalent issue, including in Europe. This problem appears to extend to textbooks used in secondary education. History education curricula and textbooks have been criticised for their Eurocentric perspective, as seen, for instance, in Austria.
Research has identified that bias among teachers, which appears to be as prevalent as in the broader population, can result in visible minority students receiving lower grades in non-blind teacher assessments, even when they perform comparably to majority students on blindly graded standardised tests, a pattern observed in several countries, including Finland, Italy, and Sweden. This bias extends to career counselling, where educators tend to be more likely to recommend lower-tier educational tracks for visible minority students with similar abilities as their majority counterparts, and are also more likely to ignore visible minority students’ counselling requests, as documented in France.
Evidence from Europe reveals that having a minority background is associated with a higher risk of being bullied at school.
Low educational attainment and hiring discrimination at entry level are major obstacles to a smooth school-to-work transition. However, visible minority youth face additional barriers that are partly driven by bias.
As shown in Germany, visible minority youth are discriminated against in access to work-based learning opportunities during formal education, such as internships and apprenticeships, limiting their ability to gain practical experience and employability.
Visible minority youth are also disproportionately subjected to disciplinary actions in school and by law enforcement, increasing dropout risks and the likelihood of a disciplinary record that deters employers. Evidence points to bias playing a role, with strong US-based research on school discipline (lacking in Europe) and studies from both the United States and Europe, notably Denmark, on law enforcement interactions.
Research conducted in a range of both EU and non-EU countries reveals that discrimination against visible minorities occurs both during and after hiring.
Correspondence studies provide compelling evidence of hiring discrimination against visible minority job candidates on both sides of the Atlantic, including in all EU countries where such studies have been conducted. These studies reveal that non-White natives, including individuals of Asian, Middle Eastern/North African, and sub-Saharan African descent, are up to twice as likely to be denied a job interview compared to their White native counterparts with equivalent CVs. Age seems to further exacerbate these disparities. Importantly, this discrimination is driven, at least in part, by bias rather than solely by employers’ expectations that visible minorities, on average, come from lower socio‑economic backgrounds, which could adversely affect their human capital – a rational economic calculation more commonly referred to as “statistical discrimination”.
Promotion bias has been documented in the United States, where Black employees face disparities despite comparable performance, though similar research is lacking for Europe. However, a major study in France reveals that bias-driven racial/ethnic discrimination in managerial supervision hinders visible minorities’ career progression – not only disadvantaging them despite similar performance but also curbing their ability to reach their full potential. Additionally, US research highlights discrimination in wage negotiations, though no equivalent studies are available in Europe.
Preliminary evidence from Germany exploring the disparate impact of the COVID‑19 pandemic on migrants and their descendants within the German labour market suggests that visible minorities may face firing discrimination – dismissal not justified by productivity differences. They are up to three times more likely to be dismissed in sectors most severely impacted by the pandemic, even when productivity is accounted for.
Evidence from the United States and Europe underscores pervasive discrimination against visible minorities in the housing market.
Field experiments reveal significant discrimination against visible minorities in both the rental and sale private housing markets in the United States and in the rental market across multiple EU countries – no study in Europe has examined the sale market. In Europe, while some differential treatment may stem from landlords’ and real estate agents’ perceptions that visible minorities, on average, come from lower socio‑economic backgrounds and may be less reliable in making regular rental payments, evidence suggests that statistical discrimination is not the sole factor at work. Bias also plays a significant role in the disparities observed between majority and minority applicants in accessing rental properties and in the price they pay, persisting even when applicants provide extensive financial information. Additionally, research in Europe indicates that bias extends to neighbours, contributing to patterns of White avoidance, where White residents tend to avoid neighbourhoods once the visible minority population surpasses a certain threshold.
While no correspondence study in Europe has examined discrimination in the private sale housing market, evidence indicates that bias limits visible minorities’ access to homeownership through discriminatory mortgage lending practices. This pattern has been documented in several European countries, including Austria, Germany, Belgium, the Netherlands, Denmark, Sweden, and Finland.
Racism can impact health through two primary channels. First, the cumulative psychological burden of repeated racist incidents can lead to deteriorating mental and subsequently physical health, affecting individuals throughout their lives. Second, bias among healthcare providers may exacerbate health disparities.
Extensive research links discrimination to poorer mental health among visible minorities, with US studies confirming a causal impact. This, in turn, can harm physical health by triggering stress pathways, increasing heart rate and blood pressure (risk factors for cardiovascular disease), elevating blood glucose levels and central fat accumulation (raising diabetes risk), and causing systemic inflammation (which may contribute to cancer). These effects can be further intensified by maladaptive coping responses, such as substance abuse and eating disorders.
Further research is needed to assess the role of healthcare providers in these disparities. Evidence on discrimination in medical appointment scheduling is mixed – confirmed in the United States but not in Germany, the only EU country where correspondence studies on this issue have been conducted. Additionally, tentative evidence from the United States suggests that bias in patient-provider interactions may be at play, although it doesn’t manifest as overt hostility towards visible minority patients but rather as lower cultural competency in dealing with them.
Visible minorities, defined as groups perceived as distinct from the majority population based on physical or cultural characteristics, make up a significant share of the population in EU countries (Box 1.1). Focusing only on a subset of these minorities – namely immigrants and their immediate descendants of non-European background, for whom data are most available – their share already ranges between 5% and 10% of the total population and is on the rise (OECD/European Commission, 2023[1]).
Yet, this subgroup faces substantial disadvantages. Some may stem from factors unrelated to racism, as non-European immigrants often come from lower-income countries and encounter greater barriers to language acquisition, recognition of foreign qualifications and citizenship compared to their peers of European descent. These challenges create disparities that often persist into the next generation.
However, racism likely plays a significant role as well, which is unacceptable from a human rights perspective. This chapter aims to provide a comprehensive overview of the available empirical evidence on the impact of racism.
Visible minorities, whether long-standing or recently arrived, are at risk of being “racialised” – not just seen as distinct, but as inherently “other” – even though this categorisation process has no biological basis, since there is only one human race (ECRI, 2021[2]). As such, racialisation reinforces the social constructs of race and ethnicity and sustains racism, “the belief that a ground such as “race”, colour, language, religion, nationality or national or ethnic origin justifies contempt for a person or a group of persons, or the notion of superiority of a person or a group of persons” (ECRI, 2017[3]), as emphasised in the EU anti-racism action plan 2020‑25 (European Commission, 2020[4]). Racism, in turn, fuels bias against visible minorities, including negative stereotypes and prejudice. When such bias influences behaviour, then it leads to bias-driven racial/ethnic discrimination – the unequal treatment of otherwise similar individuals, solely based on prejudiced and stereotypical attitudes, whether conscious or unconscious, towards people of a particular race or ethnicity.
Racialisation is not just a possibility; it is also likely, as evidence suggests that racial/ethnic bias is widespread. Racism and racial/ethnic bias are difficult to assess based on self-reported attitudes. First, such reports capture only conscious bias, failing to account for unconscious prejudice. Second, even measuring conscious bias presents challenges due to social desirability – people’s reluctance to admit socially unacceptable views. An alternative is to use the Implicit Association Test (IAT), which is designed to disclose automatic mental associations that individuals cannot easily conceal. Studies analysing IAT results reveal that one‑fourth of EU respondents exhibit a strong pro-White bias (Box 1.2). This means they are significantly slower to associate blackness with positive words like “good” or “nice” and quicker to associate blackness with negative concepts such as “bad” or “mean”, suggesting that racism may not only fuels the overtly discriminatory actions of a few individuals but also drive more subtle, often unconscious bias that can become pervasive and obstruct key areas of life (ECRI, 2017[5]).
Although racial/ethnic bias is widespread, it may not necessarily translate into bias-driven discrimination, as many individuals may successfully control their bias. However, survey-based reports of perceived discrimination suggests that bias does manifest in real-life discriminatory behaviour, either because individuals fail to fully regulate their conscious bias or because a significant share of this bias is unconscious and beyond their control. The Eurobarometer surveys from 2012, 2015, 2019, and 2023 consistently show that racial/ethnic discrimination is viewed as the most prevalent form of discrimination within the EU. In 2023, around 60% of respondents perceived discrimination based on skin colour and/or ethnic origin as frequent (European Comission, 2023[6]) – an increase of 4 percentage points since 2012. The widespread perception of racial/ethnic discrimination among the general public is corroborated by insights from racial/ethnic minorities themselves. For example, while the 2023 Eurobarometer survey indicates that 21% of the general EU‑27 population reported to have experienced discrimination over the last year, this figure rises to 36% among people of African descent, as presented in the 2023 “Being Black in the EU” report (Box 1.3). This population declares experiencing racial discrimination across various aspects of life, with conditions having deteriorated since the 2018 edition of the report.
In this report, the term “visible minorities” refers to groups perceived as distinct from the majority population based on either physical characteristics, such as skin colour, hair texture, and facial features, or cultural characteristics, including language, religion, and traditions. The former traits are commonly associated with the social construct of “race”, while the latter fall under the social construct of “ethnicity”.
The definition of visible minorities used in this report therefore differs slightly from that sometimes used in national contexts. For example, the Canadian Government defines them exclusively based on physical characteristics. According to the Employment Equity Act, visible minorities are described as “persons, other than Aboriginal peoples, who are non-Caucasian in race or non-white in colour”. While it is true that cultural attributes are not always immediately visible in the same way as physical traits, they may still be rapidly perceived through basic social interactions, and in some societies, this may lead to discrimination. A person’s name, for example, may signal cultural or religious heritage. Similarly, celebrating specific festivals can highlight cultural distinctions and attract bias. By including cultural attributes in the definition of visible minorities, a broader spectrum of ways in which individuals may be marginalised is recognised.
These preliminary findings suggest that visible minorities face discrimination from birth onwards. To effectively monitor and combat racism, a deep understanding of its consequences is crucial. This chapter examines how bias-driven racial/ethnic discrimination creates barriers across key pillars of well-being, including education, school-to-work transition, employment, housing, and health.
To ensure a robust and evidence‑based approach, the analysis prioritises empirical research over anecdotal accounts. It draws on an extensive review of social sciences literature – primarily in economics, management, political science, social psychology, and sociology – focusing on studies based on large datasets and sound impact evaluation methods. Particular attention is given to research published in top peer-reviewed journals or prestigious working paper series, as well as to analyses centred on Europe.
Before proceeding to the chapter’s main sections, four caveats are worth noting. First, our analysis focuses on bias-driven racial/ethnic discrimination, also known as taste‑based discrimination in economic literature. This means we do not examine forms of discrimination that arise from factors unrelated to bias – such as the lower socio‑economic status of visible minorities, which, as previously discussed, is partly shaped by factors beyond contemporary racism, at least in the case of recently arrived groups, i.e. immigrants and their immediate descendants of non-European background. By narrowing our focus to discrimination driven by bias, we exclude indirect discrimination, where seemingly neutral practices disproportionately disadvantage visible minorities. A key example is employee referrals, where individuals tend to recommend candidates from similar backgrounds, inadvertently excluding equally qualified individuals from disadvantaged groups who are often outside these networks. We also exclude what economists refer to as statistical discrimination, where decisions are based on risk assessment rather than bias. For instance, a landlord may reject a visible minority applicant not out of prejudice, but due to an assumption of higher financial risk. Even when applicants appear equally solvent on paper, landlords may interpret these signals as incomplete and associate visible minorities with a greater likelihood of irregular payments.
Second, we may be underestimating the full impact of bias-driven racial/ethnic discrimination by focusing only on its direct effects within each life area studied, without accounting for its broader ripple effects. Discrimination in one domain can trigger negative consequences in others. For instance, discrimination in housing can exacerbate the concentration of visible minorities in underprivileged areas, which in turn negatively affects their educational, employment opportunities, and even health outcomes, since these areas typically suffer from high levels of violence, pollution, and noise (Liebig and Spielvogel, 2021[7]). Moreover, while we attempt to capture the cumulative effects of discrimination by examining successive life stages – education, school-to-work transition, and employment – we may still fall short of fully reflecting how these disadvantages compound over time. For instance, lifelong exposure to discrimination can result in severe financial instability in old age due to inadequate retirement savings.
Third, while the focus is on Europe, many seminal studies in this field have originated in the United States, providing valuable insights that are reported in this chapter. Furthermore, in some instances, evidence is available only from the United States, requiring caution when generalizing these findings to Europe, as the histories of visible minorities in both regions differ significantly.
Fourth, while many Europe‑focused studies examine discrimination against immigrants and their immediate descendants, some extend their scope to individuals of non-European background regardless of migration history, potentially capturing visible minorities who have been settled in Europe for generations. Importantly, the limited research specifically addressing long-established visible minorities in Europe confirms that these populations also face discrimination. For instance, individuals from French overseas territories experience discrimination in the French labour market (Anne et al., 2024[8]). A study found that candidates born in Guadeloupe, Martinique, or La Réunion – bearing names typical of these regions and having completed their education and entire professional experience in mainland France – were 20% less likely to receive a callback for waiter positions in mainland France than candidates born and raised in mainland France with traditionally French-sounding names, despite submitting identical CVs. This disadvantage persists even when the situation is reversed: mainland-born candidates applying for positions in Guadeloupe, Martinique, or La Réunion still fare better, despite potentially being perceived as more of an outsider than their overseas-born peers. Similarly, extensive evidence confirms that Roma people face discrimination (Bartoš et al., 2016[9]). In Czechia, even when applicants for a rental have a college degree and stable employment, an individual with a Czech-sounding name is more than one‑third more likely to be invited for an apartment viewing than an applicant with a Roma-sounding name. This discrimination extends to the labour market. When two candidates with similar CVs – both overqualified for the position – apply for a job, the applicant with a Czech-sounding name is nearly 80% more likely to receive an interview invitation than their Roma peer.
Implicit association tests (IATs) were developed in the 1990s by social scientists Anthony Greenwald, Debbie MacGhee, and Jordan Schwartz to uncover conscious and unconscious associations between different concepts (Greenwald, McGhee and Schwartz, 1998[10]). The IAT’s most prominent application is in assessing implicit stereotypes and prejudice, revealing the bias individuals may hold, including unconsciously, regarding various racial or ethnic groups.
For instance, to measure bias against Black people, the Race IAT presents pictures of Black and White individuals alongside descriptive words on a computer screen. These words are either positive (e.g. good, pleasant, hardworking) or negative (e.g. bad, unpleasant, lazy). In the first IAT session, participants are instructed to pair pictures of White people with positive words and pictures of Black people with negative words. In the second session, the instructions are reversed: pictures of White people are paired with negative words and pictures of Black people with positive words. The IAT operates on the premise that individuals with negative bias about Black people and/or positive bias about White people will respond more quickly in the first session (associating Black people with negative words and White people with positive words) than in the second session.
Share of the population with any pro-White bias (2010‑19)
Note: The Race IAT score can take any value between ‑2 and +2. The more positive it is, the slower (resp. quicker) are individuals to associate blackness (resp. whiteness) with positive words like “good” or “nice” and the quicker (resp. slower) they are to associate blackness (resp. whiteness) with negative words such as “bad” or “mean”. An IAT score between 0.15 and 0.35 indicates a slight pro-White bias, while a score between 0.35 and 0.65 reveals a moderate pro-White bias. Scores exceeding 0.65 are considered to reflect a strong pro-White bias.
Source: Data from the Race IAT compiled on the Project Implicit website by (Coutts, 2023[11]).
IATs have been, and continue to be, administered online through the Project Implicit website. This virtual laboratory, co-founded by a team of social scientists including Anthony Greenwald, facilitates extensive research on implicit cognition. Specifically, between 2010 and 2019, nearly 3.7 million individuals took the Race Implicit Association Test. In a recent paper, economist Alexander Coutts compiled data for countries with at least 100 observations (Coutts, 2023[11]). When restricted to EU countries, his analysis reveals that, on average, nearly 7 out of 10 EU residents exhibit some form of pro-White bias, being slower at associating blackness with positive words, but quicker at associating whiteness with these concepts (Figure 1.1). While these findings do not provide evidence of the actual prevalence of racist attitudes (see for example (Meissner et al., 2019[12])), they are nevertheless a useful starting point for assessing some of the resulting bias.
Notwithstanding these caveats, it is worth emphasising two points. First, these numbers may underestimate what would be the bias if the tests were taken by a nationally representative sample of individuals in each country. Indeed, those who take the test on the Project Implicit website do so on a voluntary basis and are younger and hold more liberal political views which, according to (Coutts, 2023[11]), are two characteristics associated with lower racial bias.
Second, the IAT is considered by researchers to capture bias more accurately than explicit measures of attitudes (Bertrand and Duflo, 2017[13]). It correlates with these measures while reducing the opportunity for underreporting (see also Figure 1.2).
Relationship between Race IAT scores and the share of individuals who report being uncomfortable with having a Black son or daughter in law
Note: The Race IAT score can take any value between ‑2 and +2. The more positive it is, the slower (resp. quicker) are individuals to associate blackness (resp. whiteness) with positive words like “good” or “nice” and the quicker (resp. slower) they are to associate blackness (resp. whiteness) with negative words such as “bad” or “mean”. The share of individuals who report being uncomfortable with having a Black son or daughter in law is computed based on the following question from the 2023 Eurobarometer on discrimination: “Regardless of whether you have children or not, please tell me, using a scale from 1 to 10, how comfortable you would feel if one of your children was in a love relationship with a person from one of the following groups? “1” means that you would feel “not at all comfortable” and “10” that you would feel “totally comfortable”: A Black person”. Total “uncomfortable” is calculated by summing responses of 1 to 4, on the 1 to 10 scale.
Source: Data from the Race IAT compiled on the Project Implicit website by (Coutts, 2023[11]) and the 2023 Eurobarometer on discrimination.
According to the 2023 “Being Black in the EU” report (Box 1.3), approximately 18% (resp. 13%) of respondents who interacted with educational institutions reported experiencing racial discrimination in these settings in the past five years (resp. in the past year).
This section explores the evidence on three primary mechanisms through which racism impedes the educational trajectories of racial/ethnic minority students: quantitative and qualitative underrepresentation of visible minorities in children’s books and secondary school textbooks, bias among teachers and career counsellors, and bias among classmates.
The 2022 EU Survey on Immigrants and Descendants of Immigrants collected comprehensive data across 15 EU Member States, based on probability sampling: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Poland, Portugal, Spain, and Sweden. A total of 16 124 individuals, either originating from or with at least one parent from North Africa, sub-Saharan Africa, Syria, and Türkiye, participated. Depending on the country, the survey targeted one, two, or three specific groups. All respondents were at least 16 years old, had resided in the survey country for a minimum of 12 months, and lived in private households.
The 2023 “Being Black in the EU” report narrows its focus to a subset of these survey data, analysing responses from a sample of 6 752 immigrants and direct descendants of immigrants of African descent residing in 13 Member States: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, Poland, Portugal, Spain, and Sweden.
Source: (FRA, 2023[14]), Being black in the EU ― Experiences of people of African descent, https://fra.europa.eu/en/publication/2023/being-black-eu.
1.2.1. Bias in (text)books
This section provides an overview of the literature on bias in children’s books and textbooks.
A well-documented phenomenon in the United States
Despite advances since the seminal study of (Larrick, 1965[15]),1 evidence suggests that children’s literature in the United States continues to depict a predominantly white world. A recent analysis by a group of researchers reviewed 1 130 award-winning children’s books published in the United States from 1923 to 2019 (Adukia et al., 2023[16]). These books were categorised into two groups: “Mainstream”, recognised for their literary or artistic merit and widely used in educational settings such as schools and libraries; and “Diversity”, highlighting the experiences of underrepresented groups including women, visible minorities, and the LGBTIQ+ community.
While there has been an increase in the representation of diverse characters, the typical portrayal across both categories still features predominantly White individuals. A detailed examination of the skin tones in the illustrations of these books found a gradual shift towards including more characters with darker skin over the decades. For instance, the range of skin tones in Mainstream books from 2010‑19 closely mirrors that found in Diversity books from 1970‑79. Perhaps not surprisingly, mainstream books continue to depict lighter-skinned characters more frequently than those in the Diversity collection. Additionally, children are more often depicted with lighter skin than adults in both collections, a distinction without clear biological rationale.
The research team expanded their investigation beyond mere visual representations to also analyse how famous non-White individuals are depicted in the text of these books. In the Mainstream collection, the analysis reveals that more than 90% of the portrayed famous figures are white. The study also tracks the representation of famous individuals across different racial groups over time, in comparison to their respective population shares. Findings indicate that Black individuals and those of Latin American descent have historically been underrepresented in both the Mainstream and Diversity collections, while White individuals have been consistently overrepresented. However, the last three decades have witnessed a gradual movement toward a stronger representation of famous individuals of racial/ethnic minority background.
The researchers complement their analysis of the representation of dark skin colours and famous individuals from visible minorities by examining market dynamics that contribute to the perpetuation of marginalisation. Drawing on economic theories related to media markets, the researchers empirically test and confirm that books featuring non-dominant social identities are underproduced relative to the demand for them, due to fixed costs and other market frictions. As a result, these books are priced higher than others. On the demand side, the study utilises consumer demographics linked to book purchase data and library checkout statistics to explore consumption patterns. The data reveal a tendency among consumers to select books that reflect their own identities. Specifically, White consumers predominantly choose books with characters of lighter skin tones, whereas Black and Latin American consumers more frequently opt for books depicting darker-skinned characters. Further, by correlating local book consumption data with local consumer attitudes towards race and immigration, the researchers demonstrate a strong association between the representations in children’s books and societal views. Given that books used to educate children significantly shape their attitudes as adults, the documented patterns in children’s book purchases could explain the perpetuation of bias in books.
A phenomenon that extends beyond the United States
The issue of underrepresentation of visible minorities in children’s literature transcends national boundaries, although the evidence for Europe is very limited. In France, for example, the 40 leading publishers specializing in children’s books released, on average, just five books featuring a non-White child character over the entire decade from 2010 to 2020 – though the situation has gradually improved since 1980 (Ghelam, 2024[17]; Ghelam, 2021[18]; Thiery and Francis, 2015[19]).
Beyond mere numerical presence, the roles and activities portrayed in children’s literature are also marked by stereotyping and a downplaying of the agency and voice granted to minority characters. A study in Croatia analysing 85 children’s books used to train future preschool and primary school teachers found a relatively fair numerical presence of non-White characters, with 12 books featuring them – remarkable in a country where immigrants and their immediate descendants of non-European background make up less than 1% of the population (Butković and Vidović, 2021[20]). However, representation alone does not ensure visibility or empowerment: among these 12 books, only one features a character of colour as the protagonist.
There is also some evidence of lower representation of visible minorities in textbooks used in secondary schools. Luke Terra and Patricia Bromley conducted an analysis of 600 secondary social science textbooks to assess the incorporation of multicultural content from 1950 to 2010 (Terra and Bromley, 2012[21]). While they observed a worldwide increase in the representation of visible minorities, significant bias persists on other fronts. For example, although outdated racist or colonial terms have been removed from Finnish social science, geography, and history textbooks, such materials appear to continue to promote Western superiority, misaligning with curricular principles of equality (Mikander, 2023[22]). A similar pattern emerged in the analysis of geography textbooks in Flanders, where Western-centric perspectives prevail, reinforcing unequal portrayals of cultures (Schuermans, 2013[23]). Likewise, a review of 24 textbooks currently used in Austrian classrooms across history, political education, geography, and economic education – conducted as part of the project “Advancing Equality Within The Austrian School System” (AEWTASS) – identified bias in the representation of the African continent and its diaspora (Aping et al., 2024[24]).
In history education, curricula and textbooks have faced particularly strong criticism for promoting a Eurocentric perspective that marginalises non-European civilisations (Araújo and Maeso, 2012[25]), as evidenced, for instance, by the analysis of history textbooks in England (Nagre, 2023[26]). This marginalisation is exemplified by the scant attention given to the rich histories, achievements, and contributions of African civilisations. Additionally, history education resources frequently downplay the harmful effects of colonialism, including its role in entrenching racism and racial/ethnic bias.
The detrimental impact of bias in (text)books
The underrepresentation of visible minorities in children’s books and educational materials can hinder their educational success and broader life opportunities, by undermining their self-esteem and reinforcing negative stereotypes among the majority population.
The lack of relatable role models in educational texts can distort minority students’ view of the path from present action to future outcomes. Not seeing such examples may effectively lower their expected returns on educational efforts. If that change in expectation were then to reduce actual effort, it may lead to poorer academic performance and perpetuate a cycle of disadvantage (Delhommer, 2022[27]; Dee, 2005[28]; Marx and Roman, 2022[29]; Walker, 2001[30]). Furthermore, Eurocentric content in history textbooks might foster feelings of alienation among minority students. When educational materials neglect the historical and cultural contributions of non-European ancestors, or present a skewed narrative of their histories, students may struggle with identity and belonging. This omission can make them feel invisible and lead to the perception that their cultural heritage is undervalued or inherently inferior. Additionally, understating the detrimental impact of colonialism fails to acknowledge the historical and ongoing struggles of their communities, which can exacerbate sense of exclusion (Aasebø and Willbergh, 2022[31]).
Moreover, children’s books and textbooks play a crucial role in shaping the values and attitudes of future generations, potentially fostering an inclusive and diverse perspective among all students, regardless of their racial/ethnic background. In contrast, biased textbooks can reinforce ignorance and prejudice rather than promoting understanding and mutual respect. When youth from the majority population are not exposed to positive representations of visible minorities in children’s literature and educational materials, their perceptions, both of their own potential and that of others, can be distorted, further entrenching negative stereotypes against visible minorities. By contrast, studies have demonstrated that positive representations of members from marginalised groups can substantially lower prejudice. A notable example is Mohamed Salah, a prominently Muslim elite soccer player. A study by (Alrababa’h et al., 2021[32]) found that following Salah’s transfer to Liverpool F.C., not only did the rate of anti-Muslim tweets among the club’s fans drop by half compared to fans of other premier league clubs, but hate crimes in the Liverpool area also decreased by 16%.
1.2.2. Biased behaviours among educators
Educators may not be immune to the influence of racial/ethnic bias. In the United States, a group of researchers conducted a study to evaluate the prevalence of such bias among teachers, in comparison to the general American population (Starck et al., 2020[33]). The findings revealed that teachers exhibit levels of pro-White bias like those found in the broader population.
This section summarises the literature on the impact of bias within educational environments, focusing on whether educators – including teachers and career counsellors, who are pivotal in shaping students’ academic paths – exhibit discriminatory behaviours towards students from visible minority backgrounds.
Before proceeding, it is important to emphasise that bias among educators can lead not only to discriminatory behaviour toward students but also to unfair treatment of their parents. This consequence has been documented in the United States (see (Parsons et al., 2018[34])). Data from Europe are sparse. However, existing studies indicate that access to school for visible minority parents may be difficult. For example, a qualitative study in the northeast of England found that educational professionals often labelled South Asian parents of Bangladeshi and Pakistani heritage as “hard to reach” (Crozier and Davies, 2007[35]). Yet, the study suggests that schools also create barriers to engagement. For instance, the schools were found to apply a “one size fits all” approach, giving little recognition to parents’ needs or perspectives.
Parental involvement plays a crucial role in shaping children’s educational pathways. Students supported by proactive parental engagement are more likely to continue their education and achieve better outcomes, as supported by meta‑analyses by (Barger et al., 2019[36]; Castro et al., 2015[37]; Wilder, 2023[38]). Particularly compelling evidence comes from a large‑scale randomised control trial conducted in a socio‑economically disadvantaged district in France, with a significant share of families with immigration background (Avvisati et al., 2014[39]). The trial involved parent-school meetings aimed at increasing parental participation in their children’s education. At the end of the school year, treated families showed increased school-based and home‑based involvement activities. Consequently, students in the treatment classes exhibited more positive behaviour and attitudes at school, notably in terms of reduced truancy and fewer disciplinary sanctions.
Biased behaviour among teachers
A meta study examined the impact of racial/ethnic bias in educational settings, demonstrating that bias influences how students from visible minority groups are assessed (Malouff and Thorsteinsson, 2016[40]), although teacher bias can affect more than just grading disparities between majority and minority students. For example, evidence from the United States (where research in this area is more developed) suggests that such bias also undermines the quality of instruction provided to minority students (Jacoby-Senghor, Sinclair and Shelton, 2016[41]).
Research conducted in Europe confirms the prevalence of teacher bias in grading at the middle school level, a particularly concerning issue given that, in many European countries, middle school represents a critical juncture. At this stage, students are typically tracked into different types of high schools, shaping their future educational and career prospects.
In the United Kingdom, researchers have leveraged the dual assessment system in place for 11‑year‑olds, where student performance in English, mathematics, and science is evaluated both through a nationally set, blindly marked written exam and through assessments by their own teachers. By comparing these two evaluation methods, they identify significant disparities across racial and ethnic groups. Their findings show that Black Caribbean pupils are one‑third more likely than their White peers to receive a lower grade from their teacher than what they achieved on the national exam (Burgess and Greaves, 2013[42]).
Moreover, evidence from Italy links these disparities to teacher bias, as measured by IAT scores. Specifically, Alberto Alesina, Michela Carlana, Eliana La Ferrara, and Paolo Pinotti first confirm that students of immigrant parentage, despite performing similarly to their native peers on blindly graded standardised tests, consistently receive lower grades in non-blind teacher assessments (Alesina et al., 2024[43]). Further analysis reveals a striking pattern: teachers with higher IAT scores – indicating stronger negative bias – disproportionately downgrade high-performing immigrant students, while native students’ grades remain unaffected by teachers’ bias levels.2
Teacher grading bias is not confined to middle school; it can also affect visible minority students at both earlier and later stages of their education.
At the high school level, a field experiment in Sweden examined grading disparities in a compulsory national test (Hinnerich, Höglin and Johannesson, 2015[44]). In this study, 1 713 student tests, originally graded by the students’ own teachers, were re‑evaluated under blind grading conditions by 42 independent teachers. The results revealed a stark discrepancy: students of Swedish descent received significantly higher scores in non-blind assessments, with a difference equivalent to approximately 10% of the mean blind test score.
Similarly, in Finland, researchers analysed over half a million digital high school exit exams, leveraging a grading system where teachers initially grade student exams before they are randomly assigned to blind external evaluators, who determine the official final score (Sahlström and Silliman, 2024[45]). Their findings highlight a clear pattern: immigrant students consistently received lower scores from their teachers than from blind evaluators, particularly in subjects like literature and foreign languages, where teachers have greater discretion in grading. By contrast, grading bias was negligible in mathematics. The magnitude of this teacher grading bias was more than ten times greater than the bias observed by gender and remained significant even after accounting for students’ socio‑economic backgrounds.
Bias in teacher grading is evident even before middle school. In Germany, for instance, research has shown that primary school teachers tend to grade identical essays more harshly when they are attributed to students with Turkish-sounding names rather than German-sounding ones (Sprietsma, 2013[46]).
Despite the crucial role of early childhood education and care (ECEC) in shaping future academic and labour market success (OECD, 2018[47]), no study has yet examined bias in this critical stage of education. However, survey data reveal that fewer than two‑thirds of ECEC staff receive training on working with children from vulnerable groups (OECD, 2021[48]). While such gaps may contribute to bias, evidence suggests that they do not entirely negate the benefits of ECEC for children from immigrant backgrounds, who often face fewer opportunities to develop cognitive and socio‑emotional skills in lower-socio‑economic home environments (Heckman and Karapakula, 2019[49]; OECD, 2017[50]). In this context, ECEC proves especially advantageous for children of immigrants – by age 15, those who attended ECEC perform at a level equivalent to an additional year of schooling, whereas their peers with native‑born parents experience only half that gain (OECD, 2019[51]).
While most studies highlight teacher racial/ethnic bias in grading, one study found no evidence of such bias. In a laboratory experiment, (Van Ewijk, 2011[52]) asked 113 Dutch teachers to grade the same set of ten essays written by 11‑year‑old students, with the essays randomly assigned Dutch, Turkish, or Moroccan names. The results showed that majority teachers did not assign systematically lower or higher grades to visible minority students.
Biased behaviour among career counsellors
Racial/ethnic bias among educators can also have a profound impact on minority students’ access to career counselling. In Europe, children with immigrant parentage tend to be disproportionately steered towards vocational tracks in countries where such education is perceived as less prestigious. Conversely, in countries where vocational training is considered a solid path to the job market, these children are often underrepresented (OECD, 2017[50]). This disparity in enrolment into more academically challenging tracks is likely to significantly affect the employment prospects of these students.
Although disparities in track choice between majority and minority students could flow from differences in educational performance, recent research from Italy – where students are tracked during the transition from middle to high school – suggests otherwise (Carlana, La Ferrara and Pinotti, 2022[53]). This study demonstrates that the gap persists even when adjusting for performance. Specifically, among students of comparable abilities, as measured by standardised tests at the start of middle school, those from immigrant backgrounds are more likely to enrol in vocational rather than technical or academically oriented curricula, compared to their native peers. Notably, this trend persists across all ability levels for boys, while for girls, it appears primarily at the lower end of the ability spectrum.
The researchers suggest that these disparities may be linked to inadequate access to career counselling; notably, the gaps narrow when such counselling is provided. To reach this conclusion, they conducted a randomised control trial evaluating the “Equality of Opportunity for Immigrant Students” program, which was aimed at offering tutoring and career counselling to high-potential immigrant students. The results are striking: treated male students were 44% less likely to repeat their grade and had a 12% higher likelihood of attending academic or technical high schools, as opposed to vocational ones, compared to their peers in the control group.
Such counselling is especially crucial when bias affects teacher recommendations for prestigious academic tracks, as seen again in Italy, where teachers with stronger negative bias against immigrants are more likely to steer students with immigrant parents toward lower-tier educational paths, even when their abilities match those of their peers with native‑born parents (Carlana, Ferrara and Pinotti, 2022[54]).
Another illustrative example arises from a previously mentioned laboratory experiment involving primary school teachers in Germany (Sprietsma, 2013[46]). In this study, typical German or Turkish names were randomly assigned to a set of ten essays to evaluate the impact of perceived pupil origin on grading. As noted in the previous section, the findings confirmed that essays attributed to Turkish names received lower grades. However, the significance of this experiment extends further. It also sheds light on discriminatory behaviour concerning recommendations for secondary school tracks. In Germany, students are generally sorted into various educational tracks by the end of the 4th grade, based on teacher recommendations and parental preferences: Hauptschule, the lowest track; Realschule, the middle track; and Gymnasium, the highest track, which leads to university eligibility. Primary school teachers in the experiment were asked not only to grade the essays but also to recommend a track based on their assessment. The results demonstrated that they were 11% less likely to recommend the Gymnasium track for essays with Turkish names compared to those with German names.
Discriminatory behaviour in education counselling not only negatively affects the chances of minority students being enrolled in the most challenging secondary school tracks; it also limits their access to elite tertiary education programmes. In a comprehensive field experiment in France involving over 600 Master’s programmes and 1 800 messages sent, it was revealed that students of North African descent are over 10% less likely to receive responses from programme directors compared to their French peers when simply inquiring about application procedures (Chareyron, Erb and L’Horty, 2023[55]). Alarmingly, the programmes most likely to discriminate are those with the strongest job placement records and thus highest prestige. Consequently, unless affected candidates exert greater effort, they risk only gaining access to training with less promising professional outcomes.
The evidence presented above suggests that the academic prospects of those who are unable to access selective tracks due to discriminatory practices is considerably hindered. Supporting this observation, a natural experiment conducted in Northern Ireland in 1989 demonstrated that increasing admissions to the “elite track” (by enabling the inclusion of students who were previously on the margins of acceptance) yielded improved examination performance and higher rates of entry into higher education (Guyon, Maurin and McNally, 2012[56]).
The detrimental impact of unfair treatment by educators
Unfair treatment by educators is likely to negatively affect students’ educational attainment and achievement directly. However, the evidence also indicates a significant indirect impact, primarily through students’ internalisation of educators’ low expectations and the stereotype threat effect. These dynamics can culminate in students devaluing academic success or questioning the legitimacy of academic outcomes, leading to disengagement (Schmader, Major and Gramzow, 2001[57]). For instance, the previously mentioned study on teacher grading bias in Finland found that this bias negatively impacts visible minority students’ educational outcomes. Immigrant students with similar academic performance are nearly 20% less likely to pursue higher education if they attend a secondary school in the quartile with the highest levels of teacher bias against immigrants, compared to those in schools with a median level of bias (Sahlström and Silliman, 2024[45]).
One potential consequence is the “acting white” phenomenon, a concept originating in the United States, whereby minority students avoid academically oriented behaviours to conform to group norms (Box 1.4). However, a US-centred review of the literature presents mixed evidence regarding the existence of an oppositional culture within education among students from marginalised minority groups. On the one hand, (Fryer Jr and Torelli, 2010[58]) found that while academic achievement positively correlates with popularity for White students, this relationship is less pronounced for their Black peers. Higher grades lead to a modest increase in popularity for medium-performing Black students but negatively impact the popularity of high-performing students. On the other hand, competing research suggests that the “acting white” hypothesis is an oversimplification of minority students’ academic behaviour (Stinson, 2011[59]). Carter’s work (Carter, 2006[60]; Carter, 2005[61]) highlights the adaptability and resilience of African American students. She argues that these students do not simply reject one form of cultural capital for another; instead, they adeptly navigate both worlds. Known as “cultural navigators”, they comprehend the nuances and practicalities of both dominant and non-dominant cultural capital.
In Europe, despite limited research on the topic, evidence does not support the existence of an oppositional culture among visible minority students. In Germany, researchers examined a sample of 2 419 students across 74 secondary schools to assess the influence of peer effort, achievement, and anti-school behaviour on adolescents’ friendship choices (Lorenz, Boda and Salikutluk, 2021[62]). Results revealed that Turkish minority adolescents tend to prefer highly engaged and high-achieving peers as friends, contradicting the idea of oppositional behaviour. Similarly, in the Netherlands, a large‑scale study involving 11 215 adolescents aged 11 to 19 across 340 schools found no evidence that immigrant adolescents endorse oppositional culture to a greater or lesser extent than their majority peers (Van Tubergen and van Gaans, 2016[63]).
The concept of “acting white” gained prominence following the publication of a seminal article in Urban Review (1986) by education anthropologists Signithia Fordham and John Ogbu (Fordham and Ogbu, 1986[64]). These authors suggested that contemporary African American adolescents resist “acting white” by shunning or rejecting behaviours perceived as associated with whiteness, such as embracing the school curriculum, speaking standard English, spending significant time studying, and achieving high grades. As a result, these students inadvertently contribute to their own academic underachievement.
The concept aligns with earlier work by Ogbu (Ogbu, 1979[65]) who distinguished between “involuntary minorities”, such as those forcibly conquered (Indigenous groups) or brought to a foreign context against their will (enslaved people and their descendants), and “voluntary minorities”, or “immigrant minorities”. Ogbu argued that a long history of institutional racism has led “involuntary” minorities to believe that education offers little return in the labour market. Consequently, they develop oppositional attitudes toward education and, within their peer groups, may view certain markers of identity as belonging exclusively to the dominant racial group (Whites).
In contrast, “voluntary” minorities often perceive US society through a culturally relativistic lens, comparing opportunities in the United States to those in their homelands. This perspective shapes their response to their social, economic, and political circumstances. According to Ogbu, “voluntary” minority youth generally do not adopt oppositional identities or reject the mainstream ideology of achievement. He contends that, as a result, they tend to perform better in school and are more motivated to pursue upward mobility.
Source: (Odim and Carter, 2023[66]).
Internalisation by students of educators’ low expectations
There is substantial evidence suggesting that educators’ bias can create a self-fulfilling prophecy, leading students to internalise negative expectations and ultimately conform to them. This phenomenon, often studied in its positive form known as the Pygmalion effect, is well-documented. A notable experiment by American psychologists Robert Rosenthal and Lenore Jacobson in the late 1960s illustrated this effect (Rosenthal and Jacobson, 1968[67]). They manipulated teachers’ perceptions at the start of the academic year by falsely reporting that certain students had shown high intellectual potential on IQ tests. In fact, these students were chosen at random, and their test scores were fabricated. Nonetheless, these students later scored 50% higher on a subsequent real IQ test, suggesting that the increased attention from teachers, spurred by the belief in the students’ higher abilities, significantly boosted their performance. This experiment underscores how teachers’ perceptions, whether accurate or not, can profoundly impact students’ academic outcomes and future career opportunities.
Consistent with the internalisation of negative expectations, a study conducted in France revealed that students of low socio‑economic status (low-SES) in the Paris metropolitan area were less likely to aspire to top educational pathways compared to their peers from more advantaged backgrounds, even when their test scores were similar (Guyon and Huillery, 2021[68]). The authors find that half of this gap results from low-SES students underestimating their academic abilities. Furthermore, another 25% of the gap is because even the highest-achieving low-SES students are less informed about top educational pathways than their high-SES counterparts. Thus, the disparity in aspirations is not primarily because low-SES students place less value on top educational pathways or face higher costs. Instead, it stems from doubts of low-SES students about their own academic qualifications, and from a lack of awareness of academic opportunities echoing their hampered access to career counselling.
Accordingly, programmes aimed at building self-esteem could help mitigate disparities. The positive outcomes of the “Equality of Opportunity for Immigrant Students” program, targeted at high-potential immigrant students in Italy and mentioned previously, supports this approach. The evaluation of this initiative attributes its success to the enhancement of students’ self-confidence in their academic abilities, leading to heightened educational aspirations (Carlana, La Ferrara and Pinotti, 2022[53]).
Stereotype threat effect
Social psychologists have long recognised that individuals may experience increased psychological pressure and anxiety when they perceive a risk of confirming negative stereotypes associated with their racial/ethnic group – a phenomenon known as “stereotype threat” (Steele, 1997[69]). This increased stress can elevate blood pressure and diminish working memory capacity, potentially impairing performance (Schmader, Johns and Forbes, 2008[70]). Consequently, individuals facing stereotype threat may underperform on tests, despite having the requisite skills and knowledge to excel.
In a seminal experiment, Jeff Stone and his colleagues demonstrated that invoking a negative stereotype about a group can disorient its members (Stone et al., 1999[71]). The study leveraged the stereotype that African Americans are perceived as more athletically talented but less intellectually gifted than Whites. The researchers divided Princeton students into three groups for a sporting exercise. The first group’s activity was described as “a measure of people’s natural athletic ability”. For the second group, the same exercise was framed as “a measure of the ability to develop a strategy during a sporting performance” – implicitly, an intelligence test. The third group, serving as a control, participated in what was described simply as “sports performance measurement”. The results were telling: African Americans’ performance was lower in the second group (sports intelligence) compared to the first group (natural athletic ability), where their performance aligned with that of the control group. Conversely, White participants performed significantly worse than African Americans in the first group but outperformed them in the second. Performance levels between African Americans and Whites converged in the control group, indicating that neither group could achieve their full potential when subjected to a stigmatizing context.
Numerous additional experiments have substantiated the significant impact of stereotype threat on the academic performance of some visible minority students. This effect is notably induced by reminding students of their racial/ethnic identity prior to a test or by informing them that the test evaluates intellectual ability, as highlighted in studies by (Steele and Aronson, 1995[72]) and (Spencer, Logel and Davies, 2016[73]).
Importantly, this phenomenon is not limited to the United States, suggesting its relevance across different cultural contexts (Appel, Weber and Kronberger, 2015[74]). For instance, an experiment in Austrian schools demonstrated that the intelligence test performance of adolescents with immigrant parents declined after they were exposed to radical right election posters, while their majority peers remained unaffected (Appel, 2012[75]). Similarly, in France, students of North African descent performed worse than their peers of French descent on a verbal task – but only when it was framed as a measure of intellectual ability. When the same task was presented as unrelated to ability, their performance matched that of majority students (Chateignier et al., 2009[76]). In the same vein, research by (Sander et al., 2017[77]) revealed that migrant students in a German primary school exhibited a reduced increase in vocabulary learning when reminded that their first language was not German.
1.2.3. Bullying by peers
The 2023 “Being Black in the EU” report surveyed parents and guardians of children within the compulsory schooling age range, asking about any racist incidents their children encountered at school in the previous year. The survey focused on bullying, including verbal abuse like insults and threats, physical aggression such as hitting and hair-pulling, and social exclusion. Overall, about one in four respondents of African descent (23%) reported that their children endured offensive comments related to their ethnic or immigrant background. Additionally, around one in ten (8%) said that their children suffered from discriminatory physical abuse, and 9% indicated that their children were isolated during playtime or excluded from social events and friendship circles because of their minority status.
A cross-country study involving students in grades 4 and 8 across 11 European countries – Belgium, Cyprus, England, Hungary, Italy, Latvia, Lithuania, the Netherlands, Norway, Scotland, and Slovenia – confirms that a migration background significantly increases the likelihood of being bullied (Ammermueller, 2012[78]). This study also reveals that being a victim negatively affects both current and future academic performance. Complementary studies corroborate these findings. For instance, research from Denmark examines the determinants and effects on educational performance of being bullied at ages 10‑12 (Eriksen, Nielsen and Simonsen, 2014[79]). The findings indicate that immigrant parentage is predictive of being bullied, and that such victimisation negatively impacts students’ ninth grade GPA, with more severe effects as the intensity of bullying increases. One study focused on England and Germany, while another in the Netherlands also confirmed that racial/ethnic minority children are more likely to be victims of bullying (Wolke et al., 2010[80]; Vitoroulis and Georgiades, 2017[81]).
The long-term adverse effects of bullying can be understood within the context of General Strain Theory, which posits that individuals experiencing strain, such as bullying, may develop negative emotions like anger, frustration, depression, or anxiety (Agnew, 1992[82]). These emotions can precipitate various maladaptive responses, including wrongdoing, self-harm, or in some extreme cases even suicide. Supporting this theoretical framework, research by (Ouellet-Morin et al., 2011[83]) demonstrates that bullied children exhibit a diminished cortisol response to stress compared to their non-bullied peers, reducing their ability to cope and making them more vulnerable to anxiety, depression, emotional dysregulation, and long-term health issues.
A successful school-to-work transition, crucial for future career prospects, minimises the duration between leaving formal education and securing quality employment (OECD, 2022[84]). Conversely, a poor transition can lead to long-lasting adverse consequences, often described as the “scarring effect”. This not only sends negative signals to potential employers but also significantly raises the risk of future unemployment and lower wage (Filomena, 2023[85]). For instance, (Eriksson and Rooth, 2014[86]) demonstrate how prolonged unemployment influences employer hiring decisions in Sweden. In their correspondence study, over 8 000 fictitious applications with varied employment histories were sent to employers. The results show a clear bias against candidates with unemployment spells exceeding nine months, particularly for low and medium-skilled positions. Similarly, (Cockx and Picchio, 2013[87]) analysed the labour market outcomes of 14 660 young Belgians who remained unemployed several months after graduation. They found that scarring effects intensify with the duration of unemployment, underscoring the critical need for swift and effective transition mechanisms from education to employment.
There are several barriers to a smooth school-to-work transition for visible minority youth, including lower educational attainment and hiring discrimination in entry-level job markets. These are discussed in the previous and subsequent sections, respectively. Low attainment hampers their ability to acquire the necessary educational credentials for successful labour market entry or further education and training. As a result, these individuals are more vulnerable to unemployment, inactivity, and prolonged socio‑economic challenges. Additionally, discrimination in hiring, particularly for entry-level positions, delays their integration into the labour market and leads to skills mismatches. This can also result in poor quality employment, with visible minority students disproportionately likely to be overqualified for their first job.
This section examines two further obstacles that impede the effective transition from education to employment for racial/ethnic minority youth. The first barrier concerns their reduced access to work-based training during formal education, such as internships and apprenticeships. This hurdle impacts the breadth and depth of skills they can acquire throughout their educational journey, affecting their employability and ability to compete in the labour market. The second barrier involves their higher exposure to disciplinary actions, both in and outside schools, which increases dropout risk and the likelihood of a disciplinary record that discourages employers.
1.3.1. Reduced access to work-based learning during formal education
Strengthening work-based learning is considered essential to ensure a smooth transition from school to work, as it provides students with practical experience that enhances their employability (OECD, 2022[84]). However, visible minorities often encounter discrimination in accessing such opportunities.
Understanding work-based learning (WBL)
WBL encompasses a variety of practices conducted within workplace settings, contrasting with traditional school-based learning (Musset, 2019[88]). There are two main types of work-based learning in EU countries:
School-mediated WBL in General Education Upper Secondary Programs: this category includes internships and work placements that primarily introduce students to the world of work. Such experiences not only enhance motivation by linking classroom studies to real work contexts but also facilitate the development of soft skills and career exploration.
WBL in Vocational Upper Secondary Programs: compared to general education placements, these placements are usually longer and occur within Vocational Education and Training (VET) programmes. While some placements account for less than 50% of programme time, apprenticeships may involve longer workplace engagement. The primary goal here is for students to acquire technical skills through hands-on work experience, alongside the soft skills also developed during shorter placements. Additionally, these opportunities allow employers to assess potential future employees and provide students with insights into potential career paths.
In addition to these two types of WBL, it is important to also consider work placements within the framework of higher education. These placements can be either an official or an unofficial component of undergraduate or graduate programmes, whether in traditional universities, universities of applied sciences, or post-secondary technical colleges. These opportunities provide practical experience that complements academic learning, bridging theoretical knowledge with real-world application. Such experiences are integral in preparing students for professional environments, enhancing their employability upon graduation (Bolli, Caves and Oswald-Egg, 2021[89]; Baert et al., 2021[90]).
WBL: enhanced opportunities and challenge for visible minorities
WBL offers significant benefits, particularly for workers from disadvantaged backgrounds, including visible minorities. These groups can gain valuable insights into prospective job roles and organisational cultures, which is especially crucial for those historically subjected to workplace discrimination. This understanding can provide reassurance of fair treatment within a company, which is vital before committing to long-term employment. Additionally, WBL serves as a powerful tool to combat entrenched stereotypes in hiring practices by providing employers with firsthand exposure to candidates through internships or apprenticeships, reducing reliance on group-based bias. For example, research by (Sterling and Fernandez, 2018[91]) shows that previous internship experiences at specific employers can lead to more equitable full-time salary offers, thus helping mitigate disparities across different groups of employees. This finding underscores the role of internships in reducing employer bias by enabling direct interactions and facilitating evaluations of candidates’ actual capabilities.
While WBL offers valuable opportunities for visible minority youth, it also comes with challenges. On the one hand, where discrimination may be driven by uncertainty about actual skills, for instance in the case of immigrants with foreign education and work experience, one would expect lower differential treatment between majority and minority candidates when they apply to an internship rather than to a full-time position, due to the lower risk involved for the employer. On the other hand, visible minorities may also face stronger challenges in securing internships or apprenticeships compared to full-time positions, due to two significant demand-side mechanisms. First, candidates for internships or apprenticeships often have shorter academic and professional records, which can lead hiring decisions to be more influenced by group-level characteristics, such as race or ethnicity, than in full-time hiring scenarios. Second, firms tend to conduct less rigorous screening3 for internships or apprenticeships than for full-time roles, leaving more room for bias during the recruitment process.
Santiago Campero of the University of Toronto conducted a study to examine whether visible minorities are more susceptible to discrimination when applying for internships compared to full-time positions (Campero, 2023[92]). To investigate this, he analysed the case of a Silicon Valley software firm. Between 2009 and 2012, this company recruited for two types of positions – software engineer internships and regular, entry-level, full-time software engineer roles – that were otherwise similar across several key dimensions. The study confirms that the firm allocated less screening time to interns and that non-White candidates, including individuals of Asian, Black, and Hispanic descent, faced greater disparities in access to internships compared to full-time positions.
Confirming greater discrimination in access to WBL for visible minority students
There is evidence of discrimination against visible minorities in accessing WBL. A survey conducted in France, involving over 2000 young people seeking internships or apprenticeships, reveals that 15% reported experiencing discrimination while searching for these contracts, with those of immigrant background being disproportionately affected (Kergoat and Sulzer, 2017[93]). Additionally, a related study indicates that young people with at least one parent born outside of France are overrepresented among those facing difficulties securing internships (Farvaque, 2009[94]). The primary barriers cited include skin colour and the racial/ethnic connotations of their names, with 40% of respondents identifying each of these factors as major obstacles.
These discrepancies are not merely perceived: they reflect a real disparity. Notably, Helland and Støren (2006) analysed outcomes for over 8 000 Norwegian youths who applied for apprenticeships in 2002. They found that vocational track students of non-European origin were less likely to secure an apprenticeship than their peers with equivalent grades and school attendance (Helland and Støren, 2006[95]).
One could argue that these disparities may arise from differences in important productive characteristics that are observed by employers but not by researchers. To rule out this possibility, one must run correspondence studies. They involve sending out, in response to real job ads, the CVs and letters of application of fictitious candidates who are identical except for their racial/ethnic background, indicated by their first and/or last names. The researchers then track the number of employer responses, or “callbacks”, received by each type of applicant, with any statistically significant differences in callbacks serving as indicators of racial/ethnic discrimination. For instance, (Kaas and Manger, 2012[96]) applied this methodology in the German context by sending over 500 internship applications to firms advertising online. The only difference in the applications was the use of either a German-sounding or Turkish-sounding name. Applications with German-sounding names received 14% more callbacks. Essentially, an equally qualified applicant with a Turkish-sounding name had to send out 14% more applications to receive the same number of responses as an applicant with a German-sounding name. This result was reaffirmed in a later correspondence test focusing on internship applications within the German public sector. (Auer et al., 2022[97]) found that candidates with German-sounding names received up to 50% more callbacks compared to those with Turkish-sounding names.
These findings underscore the significant hurdles that visible minorities encounter in accessing WBL. Discrimination in obtaining internships or apprenticeships may also relegate visible minorities to lower quality internships, including unpaid or low-paid positions. While analyses regarding the types of internships secured by racial/ethnic minority students remain limited in Europe, data from the United States (National Association of Colleges and Employers, 2020[98]) confirm that Black, multi-racial, and Hispanic college students are disproportionately represented in unpaid internships and underrepresented in paid ones.
1.3.2. Greater exposure to disciplinary actions
The trend of offending behaviour peaking during the late teens and early 20s is well-documented, as shown in the seminal works of (Hirschi and Gottfredson, 1983[99]) and (Quetelet, 2003[100]). As a result, younger individuals, are more frequently subjected to disciplinary actions than older age groups.
However, this pattern is particularly pronounced among visible minority youth. Research suggests that these disparities may not only reflect a higher incidence of felonies or misdemeanours – potentially linked to lower socio‑economic status – but also racial/ethnic bias manifesting in various forms. These include disparate punitive practices within educational settings, as well as in law enforcement, including differential treatment by the police and inequities in court proceedings.
Bias in punitive practices in school
In the United States, the 2020‑21 Civil Rights Data Collection (CRDC), a comprehensive survey required of all public schools hosting students from preschool through grade 12, found that despite Black pre‑schoolers comprising only 17% of preschool enrolment, they represented a striking 31% of those subjected to one or more out-of-school suspensions and 25% of those who were expelled (U.S. Department of Education, 2023[101]).
Racial/ethnic bias appears to play a significant role in the discrepancies observed in disciplinary actions between Black and White students within school environments. As preliminary evidence, which is only correlational, a study examined the relationship between county-level explicit and implicit measures of racial/ethnic bias and racial/ethnic disciplinary disparities across approximately 96 000 schools in the United States, encompassing around 32 million White and Black students (Riddle and Sinclair, 2019[102]). The findings show a positive correlation between both explicit and implicit forms of racial/ethnic bias and Black-White gaps in five disciplinary actions, namely school arrests, expulsions, law enforcement referrals, and in-school and out-of-school suspensions. Yet, this result could stem from reverse causation, with more offenses by Black students reinforcing racial bias.
That said, evidence also suggests a tendency to impose harsher disciplinary measures on minority students, even when their behaviours are comparable to those of their peers. In a pivotal study encompassing 364 elementary and middle schools during the 2005 to 2006 academic year, (Skiba et al., 2011[103]) uncovered stark disparities: Black students were found to be twice more likely at the elementary level, and nearly four times more likely at the middle school level, to be sent to the office for behavioural misconduct compared to their White peers. Upon such referrals, Black students faced a higher likelihood of suspension or expulsion, even for actions similar to those of their White counterparts. Digging deeper into the nature of these disciplinary disparities, an earlier investigation by (Skiba et al., 2002[104]) revealed that Black middle school students were disproportionately referred for subjective, less severe infractions – such as perceived disrespect, defiance, or loitering – while White students were more often cited for objective, more serious offenses like vandalism, fighting, or drug possession.
The repercussions of such disparate treatment are profound, as they may affect both educational achievements and later-life interactions with the criminal justice system. Over time, students subjected to differential treatment may internalise a sense of being inherently “problematic” or “less capable”, leading to diminished self-esteem and lowered academic aspirations. Moreover, this uneven treatment might steer them towards affiliations with at-risk peers, perpetuating harmful stereotypes and exacerbating their marginalised status. Research corroborates these adverse effects (Bacher-Hicks, Billings and Deming, 2024[105]). The study capitalises on a significant boundary change in Charlotte‑Mecklenburg middle schools in the fall of 2002, which resulted in approximately half of the students attending a new school. The authors observe that students quasi-randomly assigned to the “stricter” schools – identified by higher suspension rates – are significantly more prone to suspension during the 2002‑03 school year, more likely to drop out of school, and less inclined to pursue college education. The ramifications extend into adulthood, with individuals assigned to stricter middle schools exhibiting higher rates of arrest and incarceration.
Comprehensive data and analyses like those available for the United States are currently lacking for Europe. Nevertheless, racial/ethnic disparities have been found in English schools, where exclusion rates for Black Caribbean students are up to six times higher than those of their White peers in some local authorities (McIntyre, Parveen and Thomas, 2021[106]).
Bias in law enforcement
Bias in law enforcement includes differential treatment by the police and uneven justice in court proceedings.
While extensive documentation exists on racial/ethnic bias in law enforcement practices in the United States, comparable research in Europe is scarce. For instance, the report “Addressing racism in policing” published in 2024 by the Fundamental Rights Agency stresses that most EU countries do not have official data sources on racist incidents and discrimination involving the police (FRA, 2024[107]). Yet, subjective measures suggest that discrimination against racial/ethnic minorities by law enforcement may be an issue beyond the United States. Notably, perceptions of discrimination within this realm are echoed on both sides of the Atlantic. In the United States, a 2019 survey by the Pew Research Center unveiled widespread perceptions of unequal treatment of Black individuals compared to Whites within the police force and the criminal justice system (Gramlich, 2019[108]). This sentiment was shared by a majority of both Black and White Americans, with Black adults approximately five times more likely than Whites to report unfair treatment by police based on their race or ethnicity. Similarly, in the EU, nearly half (48%) of respondents of African descent who had been stopped by the police in the five years preceding the survey attributed these encounters to their immigrant or racial/ethnic minority background, including factors such as skin colour or religion (FRA, 2023[14]). This figure increased to 58% among those stopped within the 12 months prior to the survey. Moreover, among those stopped, nearly one in five (19%) claimed that they were treated very or fairly disrespectfully by the police.
Bias in policing
While not necessarily evidence of actual bias and discrimination, the disproportionate likelihood of Blacks and Hispanics being subjected to police stops and searches is a well-documented phenomenon in the United States (Coviello and Persico, 2015[109]). This trend is also observed in European countries where data are available. For instance, in France, a report gathered information on over 500 police stops at five locations in and around the Gare du Nord and Châtelet-Les Halles rail stations in Paris (Jobard and Lévy, 2009[110]). The findings revealed that individuals of sub-Saharan African or Caribbean origin were six times more likely, and those of North African origin more than seven times more likely, to be stopped by the police than their White counterparts. Furthermore, the study identified a significant correlation between individuals’ ethnicity, including specific styles of clothing worn by young people, and the probability of being subjected to police stops.
Similarly, in the United Kingdom, several studies have highlighted significant racial/ethnic disparities in stop-and-search rates. Analysing 12 years of annual data from 38 police force areas in England, (Miller, 2010[111]) discovered disparities in police searches: Black individuals were 2 to 3 times more likely to be stopped and searched than Asian individuals, and 3 to 4 times more likely than those in the “others” category, which includes White people. In urban centres such as London, Manchester, and the West Midlands, the frequency for Black individuals increased, reaching nearly eight times that of others. The Lammy report, a comprehensive review commissioned by the UK Government to assess the situation of racial/ethnic minorities within the law enforcement system, corroborates these findings (Lammy, 2017[112]). According to the report, Black individuals are approximately six times more likely to be subjected to stop-and-search procedures. Recent data from the UK Government further confirm these disparities. The search-and-stop rates for Black individuals in England and Wales from March 2022 to March 2023 were reported to be 5.5 times higher than those for White individuals (Home Office, 2024[113]). Additionally, the study suggests that this disproportionality in stop-and-search activities translates into disparities in arrest rates. In 2022/23, Black individuals in England and Wales were approximately 2.2 times more likely than their White counterparts to be arrested.
The disproportionate exposure of visible minorities to police stops is especially concerning for visible minority youth. A quarter (26%) of individuals of African descent in the EU reported being stopped by the police at least once in the five years preceding the survey (FRA, 2023[14]). Among those aged 16‑24 years, the rate reaches 34%, which is three times higher than the rate for individuals aged 60 and over (11%). These findings are consistent with a 2016 survey conducted by the French equality body, which involved over 5 000 individuals (Défenseur des droits, 2017[114]). The results showed that 80% of men under the age of 25, perceived as North African/Middle Eastern or Black, reported being stopped by the police at least once in the past five years, with over one‑third experiencing more than five stops. When adjusting for factors like place of residence, education level, and financial situation, these young men were found to be 20 times more likely to be stopped than their peers of European background.
Although previous studies suggest racial/ethnic bias and discrimination, they do not provide clear-cut evidence of these mechanisms. In the United States, researchers have specifically examined this issue by investigating whether police officers discriminate against Black drivers caught speeding (Goncalves and Mello, 2021[115]). As the penalty for speeding escalates discontinuously with the speed of the driver, some officers may be inclined to mitigate the penalty by reducing the recorded speed to just below a threshold. The authors reveal that Black drivers are less likely than their White counterparts to have a reported speed just below the threshold, and document that this disparity is unlikely to stem from differences in actual speeding behaviour. Furthermore, they demonstrate that 40% of officers contribute to this discrepancy, indicating that racial/ethnic bias in policing extends beyond isolated instances. Noting the unequal rates at which Black drivers were stopped in most US counties, a subsequent study delved into the connection between racial/ethnic bias and disparities in police traffic stops (Stelter et al., 2022[116]). The authors found that the uneven stopping of Black drivers was more pronounced in counties with higher levels of anti-Black prejudice.
Evidence of racial/ethnic discrimination in policing has also been found in Europe. In Denmark, immediate descendants of immigrants are nearly 50% more likely to be arrested by the police without subsequent conviction than their Danish-born peers, even after accounting for socio‑economic factors – and this gap widens for those of non-European descent (Søndergaard and Hussein, 2022[117]).
Bias in the criminal justice system
Racial/ethnic disparities may not only materialise in the likelihood of being stopped, searched, and arrested; they can also pervade several other stages of the criminal justice system, although observed raw gaps could stem from visible minorities engaging in more severe offenses than the majority population. In the United States, Blacks are more likely than Whites to be charged with a serious crime, detained before trial, convicted of an offense, and incarcerated (Arnold, Dobbie and Hull, 2022[118]). Similar discrepancies have also been substantiated in the United Kingdom. According to the 2021 Prison Population Statistics of the UK Ministry of Justice, non-White individuals are overrepresented in British prisons compared to the general population (Ministry of Justice, 2021[119]). Despite constituting only 15% of the general population in 2020, non-White individuals made up 27% of the entire incarcerated population. Particularly striking is the fact that Black individuals account for 13% of the prison population, while they represent just 3% of the population in England and Wales. Moreover, Black individuals face not only a higher likelihood of imprisonment but also longer sentences. Since 2016, White defendants have consistently received shorter average custodial sentence lengths (ACSL) compared to defendants from other ethnic backgrounds (Ministry of Justice, 2021[119]). In 2020, the ACSL for White offenders was 19.6 months, while Black offenders received an average of 26.8 months. Asian offenders faced an average of 28.6 months, and offenders of Mixed, Chinese, or Other ethnic backgrounds received an average of 24.4 months.
These discrepancies persist even when moving away from raw gaps and comparing minority groups with their observably similar majority peers. In Spain, for example, even after controlling for gender, age, and legal factors such as offense type, African foreigners are more than four times as likely as Spanish defendants to receive a prison sentence (Riba et al., 2023[120]). Likewise, in Ireland, even after adjusting for variables such as prior custodial sentences and gender, non-Irish nationals are given longer sentences than Irish nationals for identical criminal offenses (Brandon and O’Connell, 2017[121]). Interestingly, the discrimination visible minorities experience in their interactions with the police in Denmark appears to extend into the criminal justice system. While many disparities in education and labour market outcomes between native‑born individuals with at least one foreign-born parent and those with two native‑born parents can be attributed to differences in parental socio‑economic background, one key exception stands out: even after controlling for parental characteristics, individuals with at least one foreign-born parent face a higher likelihood of receiving a prison sentence than their peers with two native‑born parents (Fjællegaard Jensen and Manning, 2025[122]).
However, these findings should not be interpreted as definitive proof of racial/ethnic bias. Disparities between visible minority groups and the majority may also stem from factors unrelated to bias, such as unequal access to quality legal assistance, often linked to lower socio‑economic status and rarely observable to researchers. To disentangle bias from both observable and unobservable factors unrelated to bias, a substantial body of literature, primarily from the United States, has developed rigorous methodologies. These studies indicate that bias does contribute to the disparities observed in the criminal justice system (Alesina and La Ferrara, 2014[123]; Arnold, Dobbie and Yang, 2018[124]; Arnold, Dobbie and Hull, 2022[118]).
For instance, Alberto Alesina and Eliana La Ferrara utilised a unique feature of the US capital sentencing process whereby all first-degree capital sentences are automatically appealed. They focused on errors made by lower courts, specifically judgments that were later overturned by higher courts. The assumption underlying their analysis was that higher courts should improve the accuracy of initial sentencing, thereby reducing or eliminating racial/ethnic bias.
A direct comparison of error rates against minority versus White defendants may be inconclusive due to potential differences in unobservables correlated with defendants’ race. However, under the assumption that these unobservables remain constant relative to the victim’s race, Alesina and La Ferrara constructed a test based on victim/defendant race pairings. Specifically, if courts are unbiased, error rates should not vary based on the race of the defendant-victim pair. For example, if courts commit more errors on minority defendants who killed White victims than on those who killed non-White victims, they should also commit more errors on White defendants who killed White victims than on those who killed non-White ones.
Analysis of an original dataset covering all capital appeals from 1973 to 1995 revealed significant racial/ethnic bias in capital sentencing: minority defendants who killed White victims were 3 to 9 percentage points more likely to face errors than those who killed minority victims, while no gap was observed between majority defendants who killed White victims relative to majority defendants who killed non-White victims. Further analysis indicated that this effect was particularly pronounced in Southern states.
According to the 2023 “Being Black in the EU” report, about one in three respondents (34%) experienced racial/ethnic discrimination when seeking employment in the five years prior to the survey, with a 12‑month prevalence of racial/ethnic discrimination at 28%. This perception extends to those who have successfully secured employment: nearly one in three respondents (31%) felt racially/ethnically discriminated against at work in the five years preceding the survey, and 23% experienced this discrimination in the past 12 months.
Drawing on the most robust empirical evidence available, this section summarises the literature on racial/ethnic discrimination across hiring, employment, and termination practices. The analysis reveals that discrimination against minorities is prevalent in hiring and employment, with some evidence indicating that this discrimination might extend to job dismissals as well.
Before proceeding, it is important to reiterate that, as highlighted in this chapter’s introduction, the following discussion – like the chapter as a whole – primarily focuses on direct discrimination. However, it should be acknowledged that indirect discrimination in the labour market likely exacerbates disparities for racial/ethnic minorities (see Box 1.5 for further discussion).
Direct discrimination occurs when an individual is treated less favourably than others under similar circumstances due to a protected attribute such as race/ethnicity, gender, age, or disability. For instance, a hiring manager who decides against employing someone because of their race/ethnicity is engaging in direct discrimination. Indirect discrimination, on the other hand, involves policies, practices, or rules that, while appearing neutral, disproportionately impact certain groups. This form of discrimination often arises unintentionally. Both forms of discrimination are unlawful in many jurisdictions. However, identifying and addressing indirect discrimination often requires a deeper analysis to uncover the broader effects of policies that may initially seem innocuous.
In the labour market, indirect discrimination mainly emanates from hiring and termination processes (Small and Pager, 2020[125]).
Regarding hiring, a common recruitment strategy involves employee referral networks, which can inadvertently perpetuate racial/ethnic uniformity. Since social circles tend to consist largely of individuals of the same racial/ethnic background (McPherson, Smith-Lovin and Cook, 2001[126]), a company with an homogenous workforce is likely to continue hiring similar profiles if relying solely on employee referrals. To mitigate this form of indirect discrimination, organisations should broaden their recruitment outreach, utilise diverse job advertisement channels, and conduct targeted job fairs and recruitment campaigns to reach underrepresented groups (OECD, 2020[127]).
Regarding termination, decisions on layoffs sometimes consider factors like tenure and/or the criticality of managerial roles, which can disproportionately affect visible minorities. This population frequently holds less tenure (specially in managerial positions) and occupies non-critical leadership role due to historical disparities (Elliott and Smith, 2004[128]). Research by (Kalev, 2014[129]) examining 327 US organisations that underwent downsizing between 1971 and 2002 found that basing layoffs on tenure or job criticality disproportionately reduced visible minority managerial representation. Alternatively, organisations that prioritised performance evaluations in their layoff decisions retained more diverse management teams. To mitigate this form of indirect discrimination, companies should consider emphasising performance over tenure or role criticality in their downsizing criteria.
1.4.1. Racial/ethnic discrimination in hiring
After examining the main findings from an extensive array of correspondence studies on racial/ethnic discrimination in hiring, this section focuses on six key additional insights drawn from this literature.
Strong evidence of hiring discrimination against racial/ethnic minorities
The first ever correspondence study was published in 1970 by Roger Jowell and Patricia Prescott-Clarke (Jowell and Prescott-Clarke, 1970[130]). The researchers compared response rates between fictitious native British applicants and fictitious applicants with similar CVs who had immigrated to Britain in their youth and held “permanent resident” status. The applicants were from four different regions: the British West Indies, Australia, Asia (India and Pakistan), and Cyprus. Their geographical origins were discernible through their names and explicit references to their home countries in cover letters. The results revealed that applicants from Australia and Cyprus achieved interview rates comparable to those of UK applicants with no migrant parentage – 74% compared to 78% for British citizens. However, applicants from racial/ethnic minorities, specifically the West Indies and Asia, faced significant discrimination, with interview rates of only 52%, approximately a third lower than those of British citizens.
Since then, hundreds of correspondence studies have been conducted to verify the prevalence of discrimination based on racial/ethnic origin. These studies overwhelmingly confirm Roger Jowell and Patricia Prescott-Clarke’s findings that visible minorities face discrimination compared to their majority peers (Lippens, Vermeiren and Baert, 2023[131]).
In a notable study, (Quillian et al., 2017[132]) performed a comprehensive meta‑analysis of 24 labour market correspondence tests conducted in the United States between 1989 and 2015. They found that, on average, White applicants received 36% more callbacks than equally qualified Black American applicants and 24% more callbacks than equally qualified Latino applicants.
Hiring discrimination against visible minorities is not confined to the United States. (Quillian et al., 2019[133]) conducted a comprehensive meta‑analysis of 97 field experiments to assess hiring discrimination across five EU countries (Belgium, France, Germany, the Netherlands, and Sweden) as well as Great Britain, Norway, the United States, and Canada. The findings revealed that non-White natives, including individuals of Asian, Middle Eastern, North African, and sub-Saharan African descent, were up to twice as likely to be denied a job interview compared to their White native counterparts with equivalent CVs. For instance, the call-back rate for White majority applicants ranged from 24% higher in Germany to 78% higher in France when compared to candidates of Middle Eastern or North African descent. Similarly, the call-back rate gap for Black applicants varied from 5% in Belgium (not statistically significant) and 49% in Great Britain to 102% in France. While there are fewer studies on hiring discrimination against Asian groups, the available data indicate significant disparities. The call-back rate gap was estimated at 35% in Norway, 42% in Canada, 45% in France, and 60% in Great Britain. Gaps in the Netherlands and the United States were smaller and not statistically significant.
The correspondence studies on which these meta‑analyses rely convey race and ethnicity through the applicants’ first and last names, which combine phenotypical and ancestry triggers of discrimination. Under these circumstances, it is challenging to discern whether phenotype (race) plays a separate role when ancestry (ethnicity) is held constant. To investigate this question, recent research utilised a large‑scale correspondence study on hiring discrimination based on racial appearance across Germany, the Netherlands, and Spain (Polavieja et al., 2023[134]). Nearly 13 000 fictitious CVs were submitted to actual job vacancies, with applicants’ phenotype and ancestry varied randomly. Phenotype was conveyed through applicants’ photographs. Specifically, the authors created eight photographs representing four phenotypic groups: Black, Asian/Indigenous, Dark-Skinned Caucasian, and White. Ancestry was signalled by ethnic-sounding names with no strong religious or class connotation, mother tongue, and parental country of origin. Regarding the latter, the study included 44 different ancestries, grouped into five main regions: Asia, Europe (together with the United States), Latin America and the Caribbean, the Middle East and North Africa, and sub-Saharan Africa. Sub-Saharan Africa was uniquely associated with Black applicants, while the other regions included applicants across all four phenotypic groups. The results provided evidence that race triggers discriminatory behaviour in all three countries. In Germany and the Netherlands, race influenced discriminatory behaviour independently of ethnicity, while in Spain, race and ethnicity jointly affected hiring outcomes.
Overall, evidence shows that racial/ethnic minorities face considerable discrimination in hiring, including due to their visibly distinct (non-White) phenotypes. This situation raises concerns about the use of artificial intelligence (AI) in recruitment processes. If these AI systems are trained on historical data, they risk perpetuating discrimination since these data are typically fraught with bias (see Box 1.6).
Machine learning algorithms rely on training datasets to build models that capture relationships between individual factors and specific outcomes. For instance, these models can predict the likelihood of an applicant being hired conditional on being interviewed, or assess an employee’s job performance once hired. Once established, the model can then be applied to forecast outcomes for new applicants.
Some observers believe that artificial intelligence (AI) can serve as a powerful tool to reduce discrimination. By uncovering predictive patterns, AI could enable employers to identify high-quality candidates that human recruiters, susceptible to bias, might overlook (Hoffman, Kahn and Li, 2018[135]). However, this benefit only holds true if AI models are trained on unbiased data; otherwise, automated methods risk reinforcing existing bias (Schellmann, 2024[136]). In an illustrative case, Bloomberg used Chat GPT to generate eight different resumes with identical educational attainment, job titles, and work experience (Yin, Alba and Nicoletti, 2024[137]). The only variation was in the names, which represented different demographic groups: men and women who were Black, White, Hispanic, or Asian. Names were randomly assigned to each resume, and Chat GPT was prompted to rank candidates for a real job opening at a Fortune 500 company. Even though all resumes were equally qualified, Chat GPT ranked one candidate highest. After 1 000 iterations with various names and combinations, the results revealed clear evidence of name‑based discrimination. Ideally, each of the eight demographic groups would be ranked as the top candidate 12.5% of the time. Instead of this, names associated with Black Americans were consistently the least likely to be selected for a financial analyst role.
Such bias is particularly likely given that most modern hiring algorithms rely on “exploitation”, which focuses on identifying the characteristics of applicants that historically predicted success. This approach, known as supervised learning, works well when firms have representative data on past applicants and when past success predictors remain relevant over time. However, these assumptions often fall short. Candidates from non-traditional backgrounds may be underrepresented in the training data particularly due to discrimination, hindering accurate performance predictions. Additionally, skill demands evolve, as demonstrated by the emphasis on remote work capabilities during and in the immediate aftermath of the COVID‑19 pandemic.
To counteract this issue, a team of researchers tackled hiring as a dynamic learning problem by assessing candidates based on their potential (Li, Raymond and Bergman, 2025[138]). Their algorithm introduces an “exploration bonus” that prioritises candidates on whom the firm has limited data. This concept, known as “hiring as exploration”, encourages firms to take a chance on lesser-known applicants. Using data from a Fortune 500 firm’s professional services recruitment, the research team demonstrated that incorporating an exploration approach into hiring algorithms improved candidate quality (as measured by eventual hiring rates) and increased demographic diversity compared to the firm’s existing practices. In contrast, algorithms based on exploitation improved hiring rates but resulted in significantly fewer Black and Hispanic applicants being selected. Specifically, the exploration approach yielded more than three times as many Black and Hispanic candidates as traditional resume‑screening algorithms.
Source: (Carcillo and Valfort, 2025 (forthcoming)[139]), Invisible Barriers. Understanding and Overcoming Discrimination in the Workplace.
Additional insights into racial/ethnic discrimination in hiring
Complementary evidence provides six additional insights into racial/ethnic discrimination in hiring. First, the real-world impact of hiring discrimination, as evidenced by field experiments, is substantial. Second, correspondence studies underestimate the true extent of discrimination against visible minorities in the labour market at large. Third, racial/ethnic hiring discrimination has not diminished over time. Fourth, age exacerbates racial/ethnic discrimination. Fifth, hiring discrimination against racial/ethnic minorities isn’t driven by a specific sector/industry, occupation, or firm’s size. Finally, racial/ethnic discrimination in hiring is driven, at least in part, by bias rather than by solely rational economic calculations.
The real-world impact of hiring discrimination, as evidenced by field experiments, is substantial
In theory, hiring discrimination may not significantly impact the labour market outcomes of visible minorities if they can compensate for it, such as by submitting more applications than their majority counterparts. However, evidence indicates that hiring discrimination does reduce employment prospects for visible minorities.
A paper published in Nature, involving a partnership with the Swiss Government-affiliated online recruitment platform Job-Room (www.job-room.ch), provides critical insights in this regard (Hangartner, Kopp and Siegenthaler, 2021[140]). Researchers employed machine learning to analyse the behaviour of employers on this platform, specifically evaluating the likelihood that employers contact “majority” and “minority” candidates with similar profiles. By linking jobseekers’ profiles to administrative unemployment data, the researchers could also examine the likelihood of “majority” and “minority” candidates securing employment, conditional on being contacted.
From March to December 2017, data were collected on 43 352 recruiters, 452 729 searches, and 17.4 million profiles that appeared in search results. The findings reveal that recruiters treat otherwise identical jobseekers differently based on their racial or ethnic background, inferred from their names, nationality, and language skills. Except for jobseekers from southern Europe, those with recent immigrant backgrounds face a significantly lower contact rate than native Swiss citizens, with penalties especially pronounced for visible minorities. Contact penalties amount to 4.2% for candidates from Western and Northern Europe, 6.2% for those from Central and Eastern Europe, 6.4% for candidates from the Americas, 12.6% for those from the Balkans, 13.5% for candidates from the Middle East and North Africa, 17.1% for those from sub-Saharan Africa, and 18.5% for individuals from Asia.
By linking jobseekers’ profiles to unemployment data, the researchers also demonstrated the real-world impact of hiring discrimination: each click on the contact button increases the likelihood that an individual exits unemployment within three months by 2.1%. In other words, jobseekers who are not contacted face a substantially reduced chance of finding employment in the same timeframe.
Correspondence studies underestimate the true extent of hiring discrimination
Correspondence studies may understate the full extent of hiring discrimination against visible minorities because they don’t assess post-interview outcomes. To test this assumption, several field experiments examining racial/ethnic hiring discrimination combined correspondence studies with audit studies. In these audits, actors represent fictitious applicants in real job interviews. Evidence from these combined studies highlights significant second-stage discrimination (Quillian, Lee and Oliver, 2020[141]): candidates representing the racial/ethnic majority not only receive approximately 50% more callbacks, but also, if invited to an interview, secure roughly 50% more job offers than minority candidates. This pattern indicates that overall hiring discrimination against visible minorities is twice as large as what is measured at the callback stage.
Racial/ethnic hiring discrimination does not seem to have diminished over time
A longitudinal analysis of correspondence studies reveals a troubling trend: racial/ethnic discrimination in hiring has not declined over time. For instance, an examination of all correspondence studies conducted in the United States from 1989 to 2015, involving fictitious White and African-American job applicants, shows that the level of discrimination against African Americans has remained consistent throughout this period (Quillian et al., 2017[132]).
Similarly, a review of correspondence studies from the late 1960s to the late 2010s across six European and North American countries (Canada, France, Germany, Great Britain, the Netherlands, and the United States) found that racial/ethnic discrimination has remained largely unchanged for most countries and visible minority groups (Quillian and Lee, 2023[142]). However, there are three notable exceptions. First, based on the available studies, it seems that hiring discrimination against individuals of Middle Eastern and North African descent increased during the 2000s compared to the 1990s. Second, discrimination in France appears to have declined, although only from very high levels to merely high ones. Third, evidence suggests that discrimination in the Netherlands has risen over time.
Finally, after synthesizing an extensive compilation of correspondence studies published between 2005 and 2020, Louis Lippens and his co‑authors found no evidence of changes in the level of hiring discrimination against racial/ethnic minorities (Lippens, Vermeiren and Baert, 2023[131]). However, caution is needed when interpreting such trends, as studies conducted over time often differ in design and typically cover only a limited segment of the labour market. As a result, identifying broad time trends remains challenging.
Age exacerbates racial/ethnic hiring discrimination
While most correspondence studies investigating racial/ethnic discrimination in hiring focus on young candidates (under 30), complementary evidence indicates that discrimination against visible minorities does not diminish with age and may even intensify.
Economist Nick Drydakis and his co‑authors, using a series of correspondence studies in the United Kingdom, explored the interplay between race/ethnicity, age, and gender (Drydakis, Paraskevopoulou and Bozani, 2022[143]; Drydakis et al., 2018[144]). They found that both White and Black fictitious applicants experienced age discrimination. In their experiments, older applicants (50‑year‑olds) possess 31 years of experience in the relevant occupation (applying for low-skilled private sector jobs, such as restaurant-café employees and sales assistants in England), while younger applicants (28‑year‑olds) only have nine years of experience in this occupation. Despite this gap, the latter were significantly more likely to receive interview invitations. Furthermore, the authors revealed that age‑related discrimination was more severe for Black applicants, particularly women.
Lastly, the study illustrated that barriers to employment access for racial/ethnic minorities not only increase with age but also that these groups are more likely to be channelled into lower-paying positions, receiving invitations for vacancies offering lower wages compared to the wages associated with the vacancies for which majority candidates are interviewed.
Hiring discrimination against racial/ethnic minorities isn’t driven by a specific sector/industry, occupation or firm’s size
Meta‑analyses of correspondence studies which assess racial/ethnic hiring discrimination control for variables such as the industry and occupation of the fictitious applicants. This approach ensures that the findings are not skewed by these factors.
Yet, correspondence studies predominantly examine private sector job listings. It is thus critical to question whether public sector employers exhibit similar discriminatory practices. While only a limited number of studies have explored this angle, the evidence generally indicates that the public sector does not fare better than the private sector in terms of equitable hiring practices (L’Horty et al., 2022[145]; Petit, Bunel and L’horty, 2020[146]; Villadsen and Wulff, 2018[147]). At best, the public sector may show slightly less discrimination, yet it is by no means free from biased hiring practices (Baert et al., 2018[148]). This may be particularly true, as research suggests that hiring discrimination in the public sector is more likely to occur at the post-interview stage, which correspondence studies fail to monitor (Cahuc et al., 2019[149]).
In addition, although meta‑analyses adjust for the occupation of the fictitious candidate, the correspondence studies they examine generally focus on non-leadership positions. To the best of our knowledge, only one correspondence study, conducted in Australia, has compared racial/ethnic discrimination in hiring for both non-leadership and leadership roles (Adamovic and Leibbrandt, 2023[150]). The findings indicate substantial discrimination at both levels, with the most severe discrimination occurring in leadership positions. Specifically, 26.8% of job applicants with English names applying to leadership roles received positive responses, compared to just 11.4% of those with non-English names – this includes 14.3% for Greek, 11.8% for Aboriginal and Torres Strait Islander, 10.3% for Chinese, 10.8% for Indian, and 9.7% for Arabic names. Thus, applicants with non-English names experienced a 15.4 percentage point lower positive response rate than those with English names. This result equates to a 57.4% lower likelihood of receiving a positive response, with the greatest penalties faced by groups whose distinctiveness is more visible. Furthermore, when applying to non-leadership roles, applicants with non-English names still face a significant though slightly lesser disadvantage; their likelihood of being invited to a job interview is 45.3% lower than that for applicants with English names.
Lastly, an examination of the influence of firm size on hiring discrimination in Belgium suggests a consistent pattern of bias. Discriminatory hiring practices are pervasive irrespective of whether a company is classified as large, medium, or small based on revenue, operating income, workforce size, and stock market listing (Baert et al., 2018[148]).
Racial/ethnic discrimination in hiring is driven, at least in part, by bias rather than by solely rational economic calculations
The results from some of the studies reviewed above might be attributed at least in part to rational economic calculations – what economists refer to as “statistical discrimination”. For instance, employers may accurately anticipate that racial/ethnic minorities, on average, come from lower socio‑economic backgrounds, which could detrimentally affect unobserved aspects of their human capital and, consequently, their productivity. As a result, employers may avoid hiring them, even in the absence of bias – a distinction from what economists call “taste‑based discrimination”.
Several correspondence studies have sought to determine whether discrimination against racial/ethnic minorities is taste‑based or statistical. Evidence for taste‑based discrimination is typically gathered using three main approaches. The first approach focuses on employer bias, assessing discrimination in relation to employers’ prejudiced views towards racial/ethnic minorities. The second approach is centred on customer contact. It involves comparing discrimination levels in jobs with high versus low customer interaction, with greater discrimination in high-contact roles suggesting taste‑based discrimination. The third approach examines co-worker contact, analysing discrimination in roles that require extensive interaction with coworkers versus those with minimal contact. Higher levels of discrimination in jobs with significant co-worker interaction indicate taste‑based discrimination.
In contrast, evidence of statistical discrimination is usually measured by introducing an experimental condition that provides additional information about candidates’ language skills, academic achievements, job qualifications, and other productivity indicators. Researchers then assess whether this additional information affects the relationship between race/ethnicity and discriminatory behaviour. If discrimination remains unchanged despite the enhanced profile, it suggests that statistical discrimination is not the primary factor. On the other hand, a decrease in discrimination upon the introduction of additional information indicates the presence of statistical discrimination.
A review of the literature indicates that both statistical and taste‑based discrimination influence hiring practices (Lippens et al., 2022[151]). Providing employers with additional information about candidates’ productivity generally reduces – but does not eliminate – their reluctance to invite racial/ethnic minority candidates to job interviews, indicating that taste‑based discrimination is at play. This is further supported by the correlation between the degree of discrimination, employers’ bias, and the level of customer and co-worker interaction required for the job.
1.4.2. Racial/ethnic discrimination while in employment
There is widespread evidence of lower wages for racial/ethnic minorities, and these persist even after adjusting for a range of observable characteristics such as education and experience. Employer-employee linked data in countries where the identification of racial/ethnic minorities is possible indicates that labour earnings disparities between majority and minority employees mainly emerge within firms rather than between them. This finding suggests that observed wage gaps are not primarily due to racial/ethnic minorities clustering in lower-paying firms.
This pattern has been documented in the United States (Carrington and Troske, 1998[152]). It has also been demonstrated in Great Britain (Forth, Theodoropoulos and Bryson, 2023[153]). Specifically, Alexander Bryson, John Forth, and Nikos Theodoropoulos identified significant racial/ethnic segregation across workplaces: approximately 60% of British firms employ no ethnic minority workers. However, this segregation does not explain the aggregate wage gap between racial/ethnic minority and White employees. Instead, most of the wage disparity exists between co-workers with comparable qualifications within the same firm. Non-White male employees earn, on average, about 10% less than their White counterparts after adjusting for wage differences across workplaces, while the wage penalty for non-White female employees is around 6%. Furthermore, the authors identified higher levels of skills mismatch and lower pay satisfaction among racial/ethnic minorities.
These findings suggest that minorities are often relegated to lower-skilled roles that do not reflect their true capabilities and/or are paid less for work of equal value. This disparity can stem from discrimination in promotion, supervision, or wage negotiation – all of which are backed by research as common forms of workplace bias. While most evidence comes from the United States, where data on this issue are more abundant, emerging research from Europe, though still limited, points to similar patterns.
Discrimination in promotion
Discrimination in promotion has been well-documented in the United Staters. Studying 9 037 new hires at a US professional services firm, a group of economists documented large racial/ethnic promotion gaps (Linos, Mobasseri and Roussille, 2025[154]): even after controlling for observable characteristics, Black employees are 18.7 percentage points (26%) less likely to be promoted than their White counterparts over the same period. Similarly, using quasi‑experimental data from a US police department, (Rim et al., 2024[155]) found that White supervisors are 28% to 40% less likely to nominate Black officers for awards than their White counterparts, even after controlling for work performance. Furthermore, the authors stress that this nomination gap widens as supervisors’ prejudice scores increase, which are measured by the use of force against Black civilians and citizen complaints.
Performance reviews appear as a critical mechanism being promotion gaps, as evidenced in the United States by a comprehensive meta‑analysis conducted by (McKay and McDaniel, 2006[156]). Another study by (Biernat and Kobrynowicz, 1997[157]) highlights how stereotypes shape evaluations of racial/ethnic minorities. Their findings reveal a double standard in assessments: while Black applicants are judged against lower minimum standards, they must meet higher ability thresholds to be considered equally qualified. This paradoxically makes it easier for these to meet basic expectations but significantly harder to be recognised as highly competent or exceptional. In alignment with these findings, (Williams et al., 2021[158]) analysed performance evaluations from a US law firm and discovered stark disparities in how employees are assessed: reviewers mentioned the mistakes of Black employees twice as often as those of White men, discussed their leadership skills 70% less frequently, and commented more on their personality traits.
The design of evaluation forms can significantly exacerbate performance review disparities. Commonly, these forms feature open-ended questions such as “Describe the employee’s accomplishments” or “How did the employee meet your expectations?”. However, the lack of specific criteria or clear definitions of expectations can introduce a high degree of subjectivity into the evaluation process. This ambiguity leaves substantial room for the evaluator’s bias to influence their judgments. Research indicates that when evaluation forms clearly define the expected competencies and require detailed evidence to support assessments, racial/ethnic disparities in reviews tend to diminish (Williams et al., 2021[158]; Mackenzie, Wehner and Correll, 2019[159]).
However, merely incorporating more quantitative measures does not automatically reduce bias. (Rivera and Tilcsik, 2019[160]) found in their quasi‑experimental study at a North American university that the structure of rating scales could significantly affect assessment bias. For instance, shifting from a 10‑point to a 6‑point scale in faculty teaching evaluations helped reduce the gender evaluation gap in the most male‑dominated fields. The study suggests that higher ratings, like a perfect 10, are often subconsciously reserved for certain groups, in this case, men, due to stereotypical associations with brilliance or high ability. A 6‑point scale, they argue, may curtail such biased perceptions.
Building on this initial finding, a recent study on a home‑services labour platform in the United States demonstrated that switching from a five‑star rating system to a simple thumbs-up/thumbs-down scale significantly reduced discrimination. Under the original system, non-White workers received lower ratings and earned only 91 cents for every dollar paid to White workers for the same job. The shift to a dichotomous scale effectively eliminated this racial/ethnic bias in customer evaluations (Botelho et al., 2025[161]).
(Bohnet, Hauser and Kristal, 2022[162]) highlight another prevalent practice in organisations in performance review process that often disadvantages racial/ethnic minorities: the use of employee self-assessments shared with managers prior to final evaluations. Their analysis of performance evaluation data from an international financial services firm reveals that some racial/ethnic minorities and women, particularly women of colour, tend to underrate themselves, possibly due to the internalisation of negative bias. These self-assessments negatively impact managers’ perceptions and decisions, creating an anchoring effect that undermines final evaluations (see (Bohnet and Chilazi, 2025[163]) for additional evidence).
(Bellé, Cantarelli and Belardinelli, 2017[164]) further substantiate the influence of such anchoring effects in an experiment involving 600 Italian public service employees. All participants received an identical description of a fictitious employee’s performance, but the experiment manipulated their expectations by varying the employee’s previous year’s performance score – 91/100 in the high-anchor group versus 51/100 in the low-anchor group. Subsequent performance ratings for the current year were significantly higher in the high-anchor group than in the low-anchor group, illustrating how initial perceptions can skew subsequent evaluations.
Discrimination in supervision
An additional factor compounds the challenges faced by visible minorities in the workplace, contributing to their disproportionately lower pay and frequent confinement to lower-skilled jobs, despite performance levels that would justify better opportunities. This mechanism involves discriminatory managerial supervision, including inadequate oversight and the assignment of tasks that are unlikely to lead to promotion.
The detrimental effects of inadequate oversight on the productivity of visible minority employees – a phenomenon known as a self-fulfilling prophecy – have been demonstrated by (Glover, Pallais and Pariente, 2017[165]). In their study, cashiers at a French grocery store chain were quasi-randomly assigned to various managers, and their performance data – including items scanned per minute and work hours – were analysed across 34 stores and 204 workers. The findings revealed that minority cashiers showed lower performance metrics, such as increased absenteeism, slower work pace, and reduced customer interaction, when supervised by biased managers identified through implicit association tests. These managers typically avoided meaningful interaction and supervision. In contrast, on days when they were managed by unbiased individuals, these minority employees outperformed their majority counterparts, underscoring the profound impact of managerial bias on worker performance.
As for bias in task assignment, evidence from the United States indicates that visible minorities are disproportionately often relegated to less favourable and less demanding tasks, which adversely affects their promotion and pay prospects, even after accounting for effort levels and career preferences (Lehmann, 2011[166]).
This pattern is not confined to the United States; there is also some – albeit very limited – evidence for Europe. A study by (Siebers and Van Gastel, 2015[167]) at the Dutch Ministry of Agriculture involved semi-structured interviews with 30 employees – half with a migration background and half without – and garnered 493 responses to an online questionnaire. The findings suggest that migrant employees were assigned less engaging tasks, diminishing their likelihood of participating actively in crucial workplace activities.
This reduced engagement may have significant implications for their earnings and career advancement, in a context where HR advisors underscore the importance of visibility for receiving positive performance evaluations and recognition as high-potential employees, echoing research by (Rodwell, Kienzle and Shadur, 1998[168]).
Discrimination in wage negotiation
The substantial racial/ethnic wage disparities between co-workers with similar qualifications within the same firm may partly be attributed to discrimination in wage negotiations.
The gender differences in negotiation behaviours have been extensively studied, revealing that traditional gender norms contribute to women’s lower self-assessment in the labour market, often resulting in lower salary requests compared to their male counterparts – a phenomenon known as the “gender ask gap”. For instance, economist Nina Roussille utilised data from an online recruitment platform for engineering roles, finding that after accounting for resume differences, there was a 2.9% gender ask gap, which ultimately contributed to a 1% pay gap (Roussille, 2024[169]).
Although research on wage negotiation dynamics related to race and ethnicity is not as well-documented as it is for gender, emerging evidence from the United States reveals influences of racial/ethnic bias on negotiation outcomes. In a 2019 study, a group of researchers conducted a salary negotiation simulation in the United States, revealing that job evaluators holding bias against Black candidates expected them to negotiate less frequently than their White counterparts, influenced by negative stereotypes about Black employees’ job performance and a prejudiced belief that they should not advocate strongly for themselves (Hernandez et al., 2019[170]). Each perceived negotiation attempt by a Black job seeker resulted in a starting salary that was, on average, USD 300 lower than the starting salary of their White counterparts. In contrast, evaluators with less racial bias offered salaries that were more equitable.
This type of bias may also be internalised by racial/ethnic minorities. For example, (Gasser, Flint and Tan, 2000[171]) demonstrated that racial/ethnic minorities tend to adopt less assertive negotiation tactics and have lower salary expectations.
1.4.3. Racial/ethnic discrimination in job dismissal
Research on racial/ethnic discrimination in job dismissals is limited but indicative of potential firing discrimination, i.e. dismissal not justified by productivity differences.
In the United States, preliminary evidence suggests that racial/ethnic minorities are more susceptible to job dismissals than their White counterparts (Wilson, 2005[172]; Hargis et al., 2006[173]), with an increased likelihood under White supervisors (Giuliano, Levine and Leonard, 2011[174]). Similarly, a study utilising Swedish population registers to compare native‑born descendants of immigrants and descendants of native‑born revealed that the disparities in unemployment rates between these two groups are primarily driven by differences in entering rather than exiting unemployment (Grotti, Aradhya and Härkönen, 2023[175]).
However, identifying the exact factors leading to job dismissals is challenging due to their often unobservable nature. Unlike correspondence studies, which can effectively – though only partially – detect hiring discrimination, experimental studies measuring the impact of discrimination in firing are impossible to implement. Therefore, it is crucial to control for variables such as industry, occupation, contract type (e.g. part-time or fixed term), and proxies for individual productivity (e.g. wage, education, and age). Indeed, these factors can significantly influence exposure to termination, independently of any discriminatory actions by employers.
In this setting, exploring firing discrimination might be most feasible during economic downturns that result in mass layoffs. Such periods provide an opportunity to compare the termination rates of majority and minority employees under more controlled and comparable conditions, helping to isolate the possible effect of discriminatory practices more effectively.
Against this backdrop, economist Daniel Auer conducted a study exploring the disparate impact of the COVID‑19 pandemic – an immense and all-encompassing economic shock – on migrants and natives within the German labour market (Auer, 2022[176]). Utilising this unexpected economic downturn as a natural experiment, Auer investigated how firms across the industrial spectrum made critical decisions on employment retention, including who to keep or place on short-time work and who to dismiss.
His findings reveal that individuals with immigrant parentage were substantially more likely to be fired compared to their native German counterparts, controlling for industry, occupation, contract type and proxies for individual productivity. Specifically, employees with immigrant parentage faced a 4 percentage point higher likelihood of job dismissal under normal industry conditions. This disadvantage escalated dramatically in sectors most severely impacted by the pandemic, with the risk of being fired for migrants rising to 24 percentage points – a firing propensity three times higher than that of persons with native‑born parents. Additionally, the study found no significant differences between the groups in terms of the likelihood to be put on short-time work.
Auer interprets these patterns as evidence that firms are more inclined to retain their workforce with native‑born parentage while demonstrating less hesitation in dismissing workers with immigrant parentage. This suggests that firing discrimination exerts additional strain on the workforce during crises. Although these findings pertain to all persons with immigrant parentage rather than visible minorities per se, they underscore the need for further research into the reality of firing discrimination against this latter group.
In the European Union, 31% of respondents of African descent reported encountering racial discrimination while trying to rent or buy property in the five years preceding the survey (FRA, 2023[14]). Nearly a quarter (23%) of these individuals declared being barred from renting by private landlords, and almost 10% said they were prevented from purchasing homes by owners or estate agencies. For example, one in ten respondents of African descent reported seeing housing advertisements that explicitly excluded or discouraged applicants with ethnic or immigrant backgrounds. Consistent with these declarations, a French real estate agency faced backlash for a blatantly racist advertisement (International Business Times, 2016[177]). The ad specified: “French nationality mandatory, no Blacks”, a discriminatory note reportedly included under the landlord’s pressure.
This instance is far from isolated. In Spain, an experimental study where researchers posed as property owners found that 72.5% of the 200 tested real estate agencies agreed to discriminate against racial/ethnic minorities when managing rentals (García Martín and Buch Sánchez, 2020[178]).
More generally, evidence underscores pervasive discrimination against racial/ethnic minorities in both the rental and sale private housing markets. Discriminatory practices extend to mortgage loan access, creating compounded barriers to homeownership for these populations.
Before delving into this literature, it is essential to understand the profound repercussions discriminatory practices in housing and mortgage lending have.
First, discrimination in the sale private housing market significantly hinders homeownership among racial/ethnic minorities. Lower homeownership rates translate into several disadvantages, including reduced economic security, housing stability, and housing quality. Homeownership is a crucial long-term investment and a primary avenue for wealth building. Moreover, homeowners generally enjoy more stability than renters, as they are less vulnerable to issues like rent increases or eviction. Additionally, homeowners have greater control over their living conditions, allowing them to invest in and improve their property, which often results in higher housing quality.
Second, discriminatory practices in the rental and sale private housing markets tend to push racial/ethnic minorities into social housing at higher rates. For instance, (Fougère et al., 2013[179]) found that from 1982 to 1999, immigrants from Türkiye, Morocco, Southeast Asia, Algeria, Tunisia, and Sub-Saharan Africa in France were significantly more likely to live in public housing. (Verdugo, 2011[180]) similarly observed that immigrants from the Maghreb were more than twice as likely to reside in public housing compared to their French native counterparts in 1999, even after accounting for factors like age, education, and family structure. The evidence reported below suggests that this pattern may not only stem from visible minorities’ lower income but also result from bias-motivated discrimination.
Third, due to capacity constraints in social housing, discrimination in the rental and sale private housing market can contribute to the fact that minorities disproportionately face homelessness (OECD, 2024[181]). In the United Kingdom, for example, they are overrepresented among the homeless even when controlling for demographics, employment, and other factors (Bramley et al., 2022[182]). Similar patterns are seen across Europe. For instance, (Baptista and Marlier, 2019[183]) reported that individuals of non-European background are overrepresented in homeless populations in Denmark and the Netherlands.
Fourth, housing market discrimination exacerbates housing segregation, the physical separation of groups into different neighbourhoods. Consistent with this mechanism, (Liebig and Spielvogel, 2021[7]) find that, across the EU, non-EU-born immigrants are more concentrated in specific areas than their EU-born counterparts. This segregation restricts access to essential resources such as quality education, healthcare, and employment opportunities, notably resulting in depressed property value. Policies that concentrate public housing in specific areas are a key factor behind the residential segregation of racial/ethnic minorities, as evidenced in Denmark and Sweden (Skifter Andersen et al., 2016[184]), as well as in France (Verdugo and Toma, 2018[185]; Verdugo, 2011[180]). However, discrimination, along with phenomena like “white flight” and “white avoidance” (see Box 1.7), also likely plays a significant role in perpetuating this segregation.
Fifth, discrimination in mortgage lending, which prevails even holding income constant, significantly impacts racial/ethnic differences in household wealth. Home equity is a major component of wealth accumulation; thus, mortgage loan discrimination reinforces the economic disparity faced by minorities.
Not only landlords and real estate agents but also neighbours may harbour bias, leading to so-called phenomena of “white flight” and “white avoidance”. Extensive literature from the United States has documented the “white flight” phenomenon, which accelerates once the minority population in a neighbourhood surpasses a certain threshold, known as the “tipping point.” (Card, Mas and Rothstein, 2008[186]) analysed US census data from 1970‑2000, finding that when the proportion of racial/ethnic minorities in a neighbourhood exceeds 5%‑20%, the White population moves out more rapidly. Historical research indicates that white flight, following the significant migration of Black individuals from the rural South to northern cities in the early twentieth century, significantly shaped the American geographic landscape, with Black populations concentrated in cities and White populations in suburbs. Economist Leah Platt Boustan estimated that each Black arrival led to 2.7 White departures (Boustan, 2010[187]). Further analysis by (Shertzer and Walsh, 2019[188]) suggests that segregation could have arisen solely from the flight behaviour of Whites.
Although less widespread, related research in Europe has emphasised a phenomenon of “white avoidance” rather than “white flight”. (Bråmå, 2006[189]) found that in Sweden, “Swedish avoidance” – characterised by low in-migration rates among Swedes to areas with increased immigrant populations – rather than high out-migration rates (“flight”), was a key driver behind the formation of immigrant concentration areas. Similarly, using longitudinal data, (Rathelot and Safi, 2013[190]) examined mobility from one French municipality (commune) to another over time and assessed the effect of the initial municipality’s ethnic composition on the probability of moving out. Their findings discredit the hypothesis of a “white flight” pattern in residential mobility dynamics in France but do reveal ethnic avoidance mechanisms among natives when relocating.
1.5.1. Racial/ethnic discrimination in the rental and sale private housing markets
After reviewing results from field experiments in the United States and Europe, this section explores whether racial/ethnic housing discrimination entirely stems from rational economic calculations or at least partly flows from bias.
Results from field experiments
In the United States, a group of researchers conducted a comprehensive meta‑analysis exploring the outcomes of field experiments to measure housing discrimination spanning the past 40 years, from the 1970s to the 2010s (Quillian, Lee and Honoré, 2020[191]). The primary source of data on discrimination in the US housing market stems from a series of large‑scale national fair housing audits conducted by the U.S. Department of Housing and Urban Development (HUD) in 1977, 1989, 2000, and 2012. These audit studies involve pairs of test applicants with different racial and ethnic identities applying for the same housing vacancies, with profiles designed to make them equally desirable renters or home buyers. These HUD studies are remarkable for their extensive scope. For example, the 2012 HUD study conducted 2 999 Black-White rental and sale tests across 26 metropolitan areas with significant African American/Black populations, including Atlanta, Chicago, Detroit, Philadelphia, and New York City.
However, the meta‑analysis does not rely solely on these audit studies. As online advertisements for dwellings have become more prevalent, audit studies have been increasingly supplemented by correspondence studies, where standardised written (e‑mail) applications are sent out, typically with only the names of applicants indicating their race or ethnicity.
The findings of this meta‑analysis reveal significant discrimination against racial/ethnic minority groups in both the rental and sales housing markets, with Black applicants facing the most severe penalties. Specifically, Black applicants have an 8 percentage point lower probability of receiving a response to an initial inquiry compared to equally qualified White applicants, while Hispanics face a 4 percentage point lower probability, and Asians a 3 percentage point lower probability. Substantial discrimination against Arab Americans is also noted, though this estimate is based on only two studies.
The analysis indicates a decline in housing discrimination from the late 1970s to the present. However, this decline is largely attributed to a reduction in overt exclusionary practices, such as not receiving a response or being told that the advertised unit is no longer available. In contrast, more subtle forms of differential treatment have remained virtually unchanged over time. In the 2010s, White auditors continued to be recommended more units and were more successful in inspecting units than their equally qualified minority counterparts.
Evidence of discrimination was also found in the private housing sector across eight European countries – France, Germany, Great Britain, Greece, Italy, Norway, Spain, and Sweden – according to a meta‑analysis focused on the rental housing market (Auspurg, Schneck and Hinz, 2019[192]), noting that no similar research has been conducted in the sale housing market. Arab/Muslim applicants consistently faced slightly higher levels of discrimination than Black applicants, while other ethnicities, primarily Central and Eastern Europeans and Asians, experienced the lowest levels of discrimination. However, the variations in discrimination across different racial and ethnic identities and countries were generally small.
Additional correspondence studies conducted after this meta‑analysis confirm significant levels of discrimination against racial/ethnic minorities. In Ireland, the first field experiment on racial/ethnic discrimination in the rental housing market demonstrated stronger discrimination against non-European migrants compared to European migrants (Gusciute, Mühlau and Layte, 2022[193]). Irish applicants were more likely to receive invitations to view apartments than Polish applicants, who were, in turn, more likely than Nigerian applicants to receive such invitations. Gender discrimination was also evident, with female applicants receiving more invitations than their male counterparts. Overall, Irish females were the most likely to be invited to view an apartment, while Nigerian males were the least likely, with a statistically significant 23 percentage point (or nearly 50%) lower response rate.
Before concluding this section, it is crucial to highlight two complementary findings from the literature on racial/ethnic discrimination in the housing market. Research suggests that differential treatment extends to the prices paid for both dwellings on sale and rental properties. For example, an analysis of 2 million repeat-sales housing transactions from four major metropolitan areas in the United States found that racial/ethnic minorities pay approximately 2% more than Whites for comparable housing (Bayer et al., 2017[194]). Similarly, (Drydakis, 2011[195]) conducted a telephone correspondence experiment in the Greek rental market, which showed that Albanian renters faced significant price disparities. Specifically, Albanian renters were offered rentals that were 1.5% more expensive in middle‑class areas and 2.3% more expensive in upper-class areas compared to their Greek counterparts.
Is bias at play?
Results from the field experiments reviewed above may be driven by statistical discrimination rather than bias. In the housing market, statistical discrimination refers to landlords and real estate agents making decisions based on accurate priors about the average characteristics of racial/ethnic minorities, which differ from those of the majority and impact profitability. One crucial characteristic is the likelihood of regular rental payments or successful bids from potential renters or buyers. Since racial/ethnic minorities have, on average, lower or more unstable household incomes than the majority population, landlords and agencies might use race or ethnicity as proxies for these unknown factors. What is more, in the case of immigrants, they often lack a credit history in the host country. In such cases, the discrimination observed may not be rooted in bias but in rational economic calculations.
Anecdotal evidence suggests that landlords often request excessive documentation or deposits from racial/ethnic minorities, particularly immigrants of African origin (Andrés Durà, 2021[196]). Moreover, field experimental evidence indicates that minority applicants who receive positive responses from landlords or real estate agents are subsequently asked for more financial information (Acolin, Bostic and Painter, 2016[197]). These are indications that statistical discrimination is at least in part at play. Additionally, the meta‑analysis by (Auspurg, Schneck and Hinz, 2019[192]) demonstrates that providing more positive information about applicants’ social backgrounds reduces discrimination.
Yet, bias is also a substantial driver of the disparities observed between majority and minority applicants in their access to property rental or sale. Significant discrimination persists in the field experiments reviewed by (Auspurg, Schneck and Hinz, 2019[192]) where extensive information on applicants’ financial status is provided.
This research aligns with findings from a laboratory experiment conducted with 576 real estate students from various higher education institutions in Belgium. The experiment shows that neither the perceived financial reliability nor the communication skills of the rental applicants significantly influence their invitation rates. In contrast, the study reveals that bias from real estate agents plays a significant role in steering minority applicants towards lower-quality dwellings (Ghekiere et al., 2022[198]).
Similarly, in the Austrian rental housing market, discrimination against applicants of foreign descent remains nearly unchanged when they state that they were born and raised in Austria – a detail that could otherwise signal integration – rather than leaving their migration background unspecified. Without providing this information, applicants with Austrian names are 10%, 33%, and 75% more likely to be invited for a viewing than those with Bosnian, Turkish, and Syrian names, respectively. Only Syrian applicants benefit from clarifying that they were born and raised in Austria, likely because it distinguishes them from Syrian refugees fleeing the civil war – a group that has faced significant prejudice (Weichselbaumer and Riess, 2024[199]).
Complementary research on the rental housing market provides compelling evidence that bias extends beyond landlords and real estate agents to neighbours, as illustrated by the phenomenon of “white flight/avoidance” (see Box 1.7). This neighbour bias has notable implications for landlords, especially those owning multiple units in the same building (building landlords), compared to those owning a single unit (dwelling landlords). If majority population members (“Whites”) avoid buildings where “Black” minorities reside, building landlords might be more likely to discriminate against Black tenants. This is because the prejudices of other tenants or prospective renters could adversely affect the landlord’s ability to rent out other units in the building. (Combes et al., 2018[200]) examined this hypothesis using data from the French National Housing Survey. Their results indicate that African immigrants in privately rented apartments are less likely to have a building landlord. Furthermore, the study reveals a positive correlation between the likelihood of African-origin tenants living in public housing and the proportion of dwellings owned by building landlords in the local housing market. These findings provide some indication that neighbour discrimination may contribute to push African tenants towards public housing.
Further research is crucial to determine whether price disparities in both home sales and rental properties are due to statistical discrimination, bias, or a combination of both. Currently, these factors are either under-researched or yield inconclusive results. For instance, the 2% premium paid by Black and Hispanic homebuyers in the United States remains even when adjusted for buyer income and access to credit, indicating a minimal role for statistical discrimination. Yet, direct evidence of bias is not strong either. Black and Hispanic buyers consistently pay more for housing regardless of the seller’s race or ethnicity, suggesting that these price disparities are unlikely to be driven by overt racial prejudice, unless sellers have prejudice against their own co‑ethnics (Bayer et al., 2017[194]).
1.5.2. Racial/ethnic discrimination in access to (mortgage) loans
Barriers to homeownership for racial/ethnic minorities are increased by discriminatory practices in mortgage loan access. Field experiments have revealed substantial discrimination by mortgage loan originators (MLOs) in their responses to loan inquiries from minorities. In the United States, for example, being African American reduces the likelihood of receiving a response from an MLO as much as having a credit score deficit of 71 points (Hanson et al., 2016[201]) – see (Ross et al., 2008[202]) for evidence of variation in such discrimination across US cities.
Similarly, a correspondence experiment by (Stefan et al., 2018[203]) in seven European countries (Austria, Germany, Belgium, the Netherlands, Denmark, Sweden, and Finland) involved sending 1 218 emails to banks. The emails were sent by individuals with either Arabic- or native‑sounding names, both with a doctorate degree to signal high educational attainment, inquiring about loans for purchasing a house. The study found that inquiries from individuals with native‑sounding names received 70% more responses than those with Arabic-sounding names (59.8% vs. 35.1%), a pattern broadly consistent across all seven countries.
Further studies have analysed two additional outcomes in the mortgage process: the likelihood of a mortgage application being denied, and the cost of the mortgage received upon approval. These analyses do not use experimental data, as it is challenging to conduct field experiments with fictitious White and minority borrowers beyond the initial inquiry phase: applying for a loan involves providing detailed, publicly verifiable financial information, such as credit scores. Consequently, these studies utilise observational data and employ the “residual method”, which examines racial and ethnic disparities in mortgage outcomes (loan approval or cost) while adjusting for non-racial borrower characteristics like income.
A meta‑analysis conducted by (Quillian, Lee and Honoré, 2020[191]), which synthesises findings from 19 observational studies in the United States, reveals evidence of racial/ethnic discrimination in both loan denial rates and mortgage costs. The analysis indicates that discrimination is more pronounced against Black borrowers compared to Hispanic borrowers, with Asians experiencing relatively low levels of discrimination. Moreover, the meta‑analysis indicates that discrimination in loan denial and mortgage costs has not significantly declined over the past 40 years, suggesting a persistent legacy of “redlining”. Through redlining, an institutional practice in the United States which was outlawed only in 1968, neighbourhoods with significant Black populations were marked in red, deeming them hazardous for investment. This practice was adopted by banks and the Federal Housing Administration, systematically denying mortgages and financial services to residents in these areas. This led to widespread racial segregation, depressed property values, and limited economic opportunities for minorities,
There is also evidence of discriminatory practices in loan denial in Europe. For instance, (Aalbers, 2007[204]) conducted in-depth interviews with real estate agents and mortgage intermediaries in the Dutch cities of Arnhem, The Hague, and Rotterdam. The findings indicate that both location and race/ethnicity significantly influence mortgage loan application assessments. Applicants who do not meet all formal criteria are more frequently approved if they are Dutch with native‑born parentage or from “low-risk” neighbourhoods, compared to visible minorities or those from “high-risk” neighbourhoods.
Research consistently shows that south-to-north international migrants tend to have lower mortality rates than the non-migrant population in their host countries – a persistent trend that can last for decades after arrival. This phenomenon is often attributed to positive health selection, meaning that migrants, on average, are healthier than the general population in their country of origin (Aldridge et al., 2018[205]; Shor and Roelfs, 2021[206]).
However, the situation for children of immigrants tells a different story. Matthew Wallace and his team have synthesised findings from several European countries, including Belgium, Denmark, the Netherlands, Norway, and Sweden (Wallace, Hiam and Aldridge, 2023[207]). These countries are distinguished by their comprehensive, register-based data systems, which reveal mortality patterns among the children of immigrants by adjusting for personal characteristics. The analysis highlights a stark divergence in mortality risks based on the geographic origin of immigrant parents. Notably, native‑born children of European immigrants generally mirror the mortality rates of their peers with native‑born parents in both early life and adulthood. However, native‑born children of non-European immigrants face higher mortality risks throughout their lives. Specifically, children with mothers from Türkiye, Somalia, and Pakistan encounter the most significant and persistent early life mortality risks. Similarly, individuals with parents from MENA countries and Sub-Saharan Africa experience higher all-cause mortality risks in adulthood. Moreover, the review highlights that while socio‑economic inequality significantly influences these health outcomes, it does not solely define them. Other critical factors, including racial/ethnic discrimination, may play a substantial role.
This section identifies two primary mechanisms through which racial/ethnic discrimination could adversely affect health. First, the cumulative psychological impact of repeated discrimination can lead to deteriorating mental and subsequently physical health, with potential life‑long and intergenerational effects. Second, inherent bias among healthcare providers may further exacerbate health disparities.
1.6.1. The impact of racial/ethnic discrimination on health
The significance of biological responses to discrimination has been largely overlooked, often due to the mistaken belief that population differences in disease risk are primarily genetic (Selvarajah et al., 2022[208]). This assumption is increasingly challenged as new research suggests that the impact of racism may affect the entire body and perpetuate a vicious cycle of harm that extends from in utero to old age, impacting individuals throughout their entire lives. It can also have intergenerational consequences via changes in maternal health.
Discrimination causes poor mental health
Extensive evidence links racial/ethnic discrimination with poorer mental health outcomes, including anxiety (Berger and Sarnyai, 2015[209]), depression (Hudson et al., 2016[210]), and suicidal thoughts and attempts (Coimbra et al., 2022[211]). For instance, a study by Stephanie Wallace and her coauthors, utilising the UK Household Longitudinal Survey, found that individuals from racial/ethnic minorities who reported experiencing racial/ethnic discrimination even once had significantly lower mental health indicators compared to those who had not experienced discrimination (Wallace, Nazroo and Bécares, 2016[212]). Moreover, for those who reportedly faced repeated instances of racial discrimination, the disparity in mental health indicators quadrupled, suggesting a cumulative impact of racism on mental well-being.
However, the relationship between reporting discrimination and experiencing mental health issues might be purely correlational, suggesting that discrimination may not per se cause psychological distress. For instance, it could be that individuals who feel discriminated against (as opposed to actual objective experience) have other conditions that may be associated with health issues.
In the United States, researchers have examined the impact of police killings of unarmed Black Americans and their spillover effects on the mental health of African Americans (Bor et al., 2018[213]). Annually, over 300 Black Americans – at least a quarter of them unarmed – are killed by police in the United States. Using a nationally representative health survey, researchers analysed differences in self-reported mental health issues among Black and White respondents exposed to one or more police killings of unarmed Black Americans in their state within three months prior to the survey. The study found that each additional police killing of an unarmed Black American was associated with 0.14 additional self-reported poor mental health days among Black respondents, while no mental health impacts were observed among White respondents. The most significant effects on mental health occurred within one to two months after exposure. Further evidence of a likely causal effect was demonstrated by the fact that mental health impacts were only observed following police killings of unarmed Black Americans, not unarmed White Americans or armed Black Americans.4
Discrimination is also associated with poor physical health conditions
Medical research finds that repeated exposure to racism overactivates stress pathways, ultimately contributing to allostatic load – a term describing the cumulative burden and “wear-and-tear” on the body and overall health (Selvarajah et al., 2022[208]). Neuroimaging studies suggest that racial discrimination may be linked to degradation in both grey and white matter of the brain. For example, in a study of 81 Black women who had experienced trauma, those who reported higher levels of racial discrimination exhibited proportionally thinner grey matter in the cingulate cortices (Fani et al., 2022[214]). Additionally, racial discrimination compromises the integrity of white matter in the prefrontal cortex (Okeke et al., 2023[215]). These brain regions are not only crucial for emotional regulation and thus mental health, they also play a key role in maintaining neurological functions.
Moreover, evidence suggests that allostatic load results in increased heart rate and blood pressure (Selvarajah et al., 2022[208]), which are risk factors for cardiovascular diseases. It also leads to elevated blood glucose levels and central fat accumulation (Butler et al., 2002[216]), increasing the risk of diabetes, as well as systemic inflammation, particularly in the gut and microbiome (Dong et al., 2023[217]), which can contribute to cancer risk. The impact of racism on these outcomes may extend beyond physiological effects; it can also manifest through behavioural responses. Maladaptive coping mechanisms, such as substance abuse or eating disorders, can further exacerbate health issues. For instance, individuals facing discrimination may engage in emotional eating (Raney et al., 2023[218]) or turn to drugs (Carter et al., 2019[219]), behaviours that increase risks of health problems like cardiovascular disease, diabetes, and metabolic problems.
Finally, allostatic load can trigger epigenetic responses, where environmental factors modify gene expression. For example, the cumulative strain on the body experienced by discriminated populations has been linked to epigenetic ageing – when biological age surpasses chronological age. This phenomenon is a predictor of coronary heart disease, diabetes, other age‑related chronic illnesses, and premature mortality. Notably, a recent study published in the Journal of Racial and Ethnic Health Disparities explored the link between interpersonal racism and gene expression regulation – specifically DNA methylation – from blood samples of 384 participants in the Black Women’s Health Study (Ruiz-Narváez et al., 2024[220]). Researchers used five different epigenetic clocks to assess ageing in participants, adjusting for factors such as age, body mass index, and socio‑economic status. They identified significant changes in DNA methylation associated with reported experiences of racism, particularly among those who reported racism in their daily lives.
Discrimination can impact individuals’ health throughout their lives
Racism may negatively impact health, from in utero to end of life (Gee, Walsemann and Brondolo, 2012[221]).
The pre‑, peri- and postnatal period
Chronic racial/ethnic discrimination is linked to significant changes in maternal biology during pregnancy (Chaney et al., 2019[222]). Such stress can lead to higher rates of birth complications, maternal mortality, excessive weight gain, and deteriorating physical and mental health. These adverse effects can extend to their offspring, resulting in increased rates of preterm birth, low birthweight, and congenital anomalies (Alhusen et al., 2016[223]; Sheikh et al., 2022[224]).
A study utilising comprehensive administrative data from California birth records, hospitalisations, death records, and parental income from Internal Revenue Service tax records has provided new insights into racial and ethnic disparities in maternal and infant health (Kennedy-Moulton et al., 2025[225]). The outcomes for Black families at the top of the income distribution are significantly worse than those for White families at the bottom of the income distribution. The maternal mortality rate for Black mothers in the top income quintile stands at 4.3 deaths per 10 000, which is approximately 60% higher than the rate of 2.7 deaths per 10 000 among White mothers in the bottom quintile. Additionally, infants born to Black parents in the highest income bracket have low birth weight and preterm birth rates that are 1.5 to 2 times higher than those of infants born to White parents in the lowest income bracket. Specifically, 14% of Black infants in the top income ventile are low birth weight, and 15% are preterm, compared to 7.4% and 9.1%, respectively, for White infants in the bottom ventile. Furthermore, the infant mortality rate for Black infants in the top decile of the income distribution is 6.4 deaths per 1 000 births – around 10% higher than the rate of 5.8 deaths per 1 000 births among White infants in the bottom decile.
Yet, these residual gaps, which persist even after adjusting for patients’ socio‑economic characteristics, may be driven by several factors other than discrimination. Such factors include, for instance, differences in health-seeking behaviours between majority and minority patients.
Childhood, adolescence and adulthood
As children and adolescents engage more with society outside their family, they face increasing social interactions that may heighten their awareness of their marginalised status (Spears Brown and Bigler, 2005[226]). As seen above, discrimination is embedded within numerous institutions that influence these formative years, particularly in the education system. Such entrenched discrimination can have profound and long-lasting effects on health and well-being.
The transition from adolescence to adulthood is a pivotal stage, often marked by significant life events which can be influenced by discrimination, such as unemployment, educational failure or even incarceration (Gee, Walsemann and Brondolo, 2012[221]). By adulthood, the cumulative effects of discrimination can manifest after latency as overt mental and physical health problems.
Old age
In later life, the cumulative impact of exposure to discrimination may exacerbate the biological effects of ageing, leading to significant physical and mental health issues that often result in irreversible comorbidities, such as reduced hippocampal volume and impaired memory (Forrester et al., 2019[227]). Continued exposure to stress is also linked to chronic degeneration, atrophy, and impaired neuronal function in the prefrontal cortex, further compromising cognitive abilities. For instance, Black older adults in the United States are nearly twice as likely to develop Alzheimer’s disease or other forms of dementia compared to their White counterparts (Alzheimer's Association, 2024[228]), although other factors than racial discrimination may explain these disparities.
1.6.2. Bias in the healthcare system
The detrimental impact of discrimination on the health outcomes of visible minorities may be further magnified by bias within healthcare systems. A comprehensive international survey conducted by SANOFI in 2022 and 2023 involved over 24 000 individuals across ten countries – Australia, Brazil, Canada, France, Germany, Japan, Mexico, Spain, the United Kingdom, and the United States (Sanofi, 2024[229]). The findings highlight that racial/ethnic minorities, alongside other marginalised groups such as women, LGBTIQ+ individuals, and people with disabilities, consistently report worse healthcare experiences. Common grievances include not being heard, feeling unwelcome, facing judgment, and in some instances, feeling unsafe within healthcare environments.
Consistent with these findings, the 2023 report “Being Black in the EU” reveals that 9% of respondents of African descent reported facing racial discrimination in healthcare settings in the year prior to the survey (FRA, 2023[14]). This included negative interactions during visits to healthcare professionals such as doctors, nurses, and dentists, and in various settings including hospitals, emergency clinics, and medical centres.
Research by (Rivenbark and Ichou, 2020[230]) used data from a nationally representative cross-sectional survey conducted in France between 2008 and 2009, which included 21 761 participants. Their study reveals that racial/ethnic minorities, particularly those from Sub-Saharan or North African descent, reported significantly higher rates of discrimination in healthcare settings, even after adjusting for key factors such as socio‑economic status. (Hanssens et al., 2017[231]) come to the same conclusion after analysing survey data among 61 932 patients from 30 European countries. This (self-reported) discrimination is associated with detrimental health behaviours: racial/ethnic minorities reporting experiences of discrimination were significantly more likely to skip necessary medical care. After adjusting for age, marital status, socio‑economic background, and health status, (Rivenbark and Ichou, 2020[230]) show that self-reported discrimination is associated with a 14 percentage point increase in the predicted probability of foregoing care.
This section first explores the available evidence regarding possible bias against racial/ethnic minorities among healthcare professionals. It then investigates whether these minorities may receive differential treatment, notably drawing on field experiments conducted in real-world healthcare settings.
Evidence of racial/ethnic bias among healthcare professionals
Extensive research, largely from the United States, suggests that healthcare professionals, including nurses, personal care workers, and doctors, display levels of pro-White bias similar to those found in broader society (Maina et al., 2018[232]; FitzGerald and Hurst, 2017[233]; Hall et al., 2015[234]). This bias, detected using implicit association tests, echoes other forms of prejudice. For instance, a study has found that racial/ethnic disparities in pain management – where Blacks are systematically undertreated for pain relative to Whites – may be associated with negative stereotypes against racial/ethnic minorities (Hoffman et al., 2016[235]). Specifically, this study reveals that half of a sample of more than 400 White medical students and residents harbour erroneous beliefs about biological differences between Black and White individuals, such as the myth that Black people’s skin is thicker than White people’s skin. These misconceptions influence clinical judgments. Specifically, those endorsing these beliefs tend to underestimate the pain experienced by Black patients and make less accurate treatment recommendations. Similarly, the belief that visible minorities exaggerate pain contributes to their undertreatment. This bias is even reflected in informal medical terminology, with some European doctors using terms like “morbus mediterraneus” or “morbus Bosporus” to characterise what they perceive as heightened pain expression among visible minority groups (von Eisenhart Rothe, 2024[236]).
The susceptibility to bias in healthcare is further exacerbated by insufficient knowledge about how symptoms and conditions manifest on skin of colour, often resulting in misdiagnoses and eroded patient trust. Most medical textbooks predominantly depict conditions as they appear on lighter skin (Louie and Wilkes, 2018[237]). While not discriminatory, this may lead to inadequate treatment for visible minorities as many symptoms – from simple rashes to more complex diseases like Neurodermitis – appear differently on darker skin. For example, the typical red rings indicative of Borreliosis on white skin present as bluish-grey on black skin. Such discrepancies in clinical recognition can lead to severe outcomes, including nerve inflammation from untreated Borreliosis. In response to this critical gap in medical education, Mukwende, Tamony, and Turner published a textbook in 2021 in the United Kingdom that illustrates how various symptoms manifest across different Black and Brown skins, thereby fostering a more inclusive and accurate approach to medical diagnosis (Mukwende, Tamony and Turner, 2020[238]).
Limited evidence that bias causes differential treatment in healthcare settings
Meta‑analyses reveal a positive correlation between racial/ethnic bias in healthcare providers, as measured by the implicit association test, and outcomes for racial/ethnic minorities (Maina et al., 2018[232]; FitzGerald and Hurst, 2017[233]; Hall et al., 2015[234]). These studies specifically indicate that increased health provider bias is associated with poorer patient-provider interactions, inadequate treatment decisions, lower treatment adherence, and ultimately, worse patient health outcomes.
A scoping review of (mostly qualitative) studies focused on migrants in Europe reaches a similar conclusion. Both overt and covert displays of racism by healthcare providers are associated with compromised quality of diagnosis and treatment (Pattillo et al., 2023[239]).
However, these relationships cannot be deemed causal, as they may be influenced by various confounding factors. For example, the quality of the provider could play a role, if lower quality correlates with stronger bias. Moreover, residual gaps in health outcomes, even after adjusting for patients’ socio‑economic characteristics and comorbidities, do not conclusively indicate discrimination. A case in point is (Graham, 2016[240])’s study on racial/ethnic differences in two cardiovascular diseases – acute coronary syndrome and myocardial infarction – in the United States, which demonstrated that Black and Hispanic patients experienced longer treatment delays and poorer outcomes than their non-Hispanic White counterparts. However, these disparities do not inherently prove causation, as there may be unobserved variables correlated with different racial/ethnic identity, such as differences in health-seeking behaviours, that influence health outcomes independently of any discriminatory practices by health providers.
Field experiments, which control for various factors to isolate the effect of racial/ethnic identity, are invaluable for examining differential attitudes by healthcare providers. These studies are particularly effective for analysing disparities in access to healthcare. However, applying them to investigate disparities in patient-provider interactions and the care received by visible minority and majority patients is more challenging.
Mixed evidence that bias leads to differential access to healthcare
Only two field experiments have examined bias in medical appointment requests, carefully ensuring that all fictitious patients had identical insurance to eliminate financial risk as a factor. One study focuses on the mental health sector in the United States, while the other investigates non-mental health care in Germany. The findings are mixed, revealing evidence of discrimination in the United States but not in Germany.
In the United States, researchers manipulated three attributes of help-seekers: social class (middle or working), gender (female or male), and race (Black or White) (Kugelmass, 2016[241]). Using voice‑over artists, they recorded messages that portrayed racially distinctive names and adopted race‑ and class-specific speech patterns. These recordings, left on the voicemail of 320 therapists overnight, described symptoms of depression or anxiety, mentioned the same health insurance plan, and requested a callback with available appointment slots. The study found that White middle‑class help seekers are more than 50% more likely (28% vs. 17%) than their Black counterparts to receive an appointment offer from therapists, with disparities being more pronounced for men (28% vs. 13%) than for women (29% vs. 21%). Conversely, the lower callback rates among working-class callers did not significantly vary by race or gender.
At first glance, these discrepancies could be driven by statistical discrimination, as financial concerns surely influence therapists’ decisions. This might lead them to discriminate against both working-class and middle‑class Black individuals, assuming a greater financial risk. However, the design of the experiment – which targeted therapists within the same insurance network covering all help-seekers involved in the experiment – was specifically chosen to eliminate any “rational calculus” based on financial risk.5
It is worth noting that a second field experiment was conducted in the US mental health sector, though it was not designed to rule out statistical discrimination as a potential factor (Fumarco et al., 2024[242]). Using a popular online platform, sent emails to an array of mental health providers, including psychologists, counsellors, social workers, and psychiatrists. These emails requested appointments for common mental health issues such as anxiety, depression, and stress. The experiment manipulated three attributes of the help-seekers: gender (female or male), race/ethnicity (suggested through carefully chosen first and last names to imply middle, not low, socio‑economic status), and transgender/non-binary identity. Specifically, for transgender/non-binary prospective patients, each request included the following statement: “I am (a transgender woman)/(a transgender man)/(non-binary) and am looking for a trans-friendly therapist”. The findings show that cisgender White patients were significantly more likely than African American and Hispanic patients to receive an appointment or a call offer. This discrepancy was particularly pronounced among transgender or non-binary individuals from these visible minority groups. That said, the experimental setup did not account for whether the help-seekers had health insurance, a factor that could influence the results, especially given that transgender and non-binary individuals are statistically less likely to be insured and tend to have lower incomes.
While a field experiment in the United States suggests that bias, rather than purely rational decision-making, plays a significant role in the unequal access to healthcare faced by visible minorities, no such evidence was found in Germany. In a large‑scale email correspondence study, fictitious patients with German- and Turkish-sounding names (each with identical health insurance) requested appointments from over 3 000 physician offices across the 79 largest German cities. The study, which targeted dentists, ophthalmologists, dermatologists, and orthopaedists, found no evidence that perceived German or Turkish descent influenced the likelihood of securing an appointment or the length of wait times (Halla, Kah and Sausgruber, 2021[243]).
The differing findings between the United States and German studies may be partly explained by study design, underscoring the need for further research. The German study assured practitioners that only a single appointment was required, potentially making interactions with visible minority patients less costly for biased doctors. In contrast, the US study involved mental health providers, who likely anticipated ongoing interactions with patients, which may have created conditions that allowed bias to influence appointment scheduling.
Tentative evidence that bias may impact patient-provider interactions
Field experiments on healthcare access provide only a partial view, as they do not account for potential discrimination in critical areas such as diagnosis, billing, or treatment. However, no field experiment has yet directly compared how White practitioners interact with patients across different racial/ethnic backgrounds under controlled, equal conditions.
That said, a field experiment in Oakland, California, suggests that bias may be at play – not as overt hostility toward visible minority patients, but rather as lower cultural competency among White doctors when treating non-White patients (Alsan, Garrick and Graziani, 2019[244]). Specifically, the study randomised more than 600 Black men to either Black or non-Black male doctors for a consultation about preventive cardiovascular screenings. Black patients paired with Black doctors were 18 percentage points more likely to agree to preventive procedures after consultation, compared to those with non-Black doctors. This trend was even more pronounced for invasive tests, such as diabetes and cholesterol screenings, which require a blood sample and more heavily depend on patient trust compared to non-invasive measures like body mass index and blood pressure measurements.
This outcome doesn’t seem to be driven by outright tension between patients and practitioners of different racial/ethnic backgrounds. Before the consultation, patients could choose which screenings they wanted via a tablet that introduced their doctor with text and photos. If there was a bias against doctors of different racial/ethnic backgrounds, it would likely appear at this stage. However, the choice of preventive measures was similar regardless of the doctor’s race/ethnicity, indicating no immediate prejudice based on the doctor’s photo. Additionally, feedback from Black patients’ post-consultation was equally positive for both Black and White doctors.
The critical insight from this study is that shared racial/ethnic backgrounds might facilitate better communication, a vital element in clinical care where effective information exchange can be lifesaving. Observations from the clinical interactions (as reflected by the doctors’ notes) revealed that patients were more forthcoming and sought more advice when their doctor was Black. This phenomenon may reflect subtle (and unconscious) expression of bias among White doctors, manifesting as lower cultural competency when interacting with non-White patients.
Although not rooted in overt racism, the lower quality of interactions between doctors and patients of different racial/ethnic backgrounds has profound repercussions. The authors estimate that improving screening uptake through enhanced doctor-patient interactions for Black patients – achieved either through racial/ethnic concordance or improved cultural competency among White doctors – could reduce the cardiovascular mortality gap between Black and White men by 19% and the life expectancy gap by 8%.
This comprehensive review of academic research reveals that bias-driven racial/ethnic discrimination obstructs key pillars of well-being, including education, school-to-work transition, employment, housing, and health.
In education, bias manifests in three ways: underrepresentation in children’s books and secondary school textbooks, bias among teachers and career counsellors, and bias among classmates – each with strong potential or clear evidence of negative effects on the educational trajectories of visible minority students.
In school-to-work transition, visible minority youth face discrimination in accessing work-based learning opportunities during formal education, limiting their ability to gain practical experience and reducing their employability. They are also disproportionately subjected to disciplinary actions, both in school and by law enforcement, with evidence that bias contributes to these disparities, increasing dropout risks and the likelihood of a disciplinary record that deters employers.
In employment, bias-driven racial/ethnic discrimination occurs both during and after hiring. A wealth of correspondence studies confirms significant hiring discrimination against non-White applicants, with gaps persisting even when fictitious candidates provide strong signals of employability and productivity in their CVs, indicating that bias, not just risk assessment, is at play. Discrimination also hampers visible minorities’ career advancement – not only disadvantaging them despite similar performance but also limiting their ability to reach their full potential. Moreover, tentative evidence from Germany suggests that visible minorities may also face firing discrimination.
Bias-driven racial/ethnic discrimination is also prevalent in housing, particularly in the private rental market. As in employment, bias plays a key role, with significant discrimination persisting even when rental applicants provide extensive financial information. While no correspondence study in Europe has examined discrimination in private home sales, evidence suggests that bias hinders visible minorities’ access to homeownership through discrimination in mortgage lending.
In health, extensive research links discrimination to poorer mental health among visible minorities, with US studies confirming a causal impact. This, in turn, can harm physical health by triggering stress pathways, effects that may be further compounded by maladaptive coping responses, such as substance abuse and eating disorders. Bias among healthcare providers could exacerbate these impacts, but further research is needed to confirm this.
A substantial share of the evidence presented comes not only from the United States but also from a wide range of European countries, which is crucial for ensuring the relevance of this literature review to EU countries, given the markedly different histories of visible minorities in the United States and Europe. The literature review also incorporates the latest research to ensure that recent progress – such as improvements in the integration of immigrants and their native‑born descendants in Europe (OECD/European Commission, 2023[1]) and growing efforts to combat racism and discrimination – is at least partly captured in the analysis.
However, ongoing literature reviews focusing on newly released research are needed to more fully reflect these developments. Moreover, significant gaps persist in key areas. For instance, evidence on disparities in disciplinary actions in schools, interactions with law enforcement, access to healthcare and treatment by healthcare providers remains limited or entirely absent in Europe, highlighting the need for further research.
That said, for many mechanisms examined, recent European studies already reveal worrisome patterns – especially given that bias-driven racial/ethnic discrimination in one life area often amplifies disadvantages in others, accumulating both across life domains and over time. These findings call for urgent policy action, starting with the development of a robust framework to monitor and assess the impact of racism and the effectiveness of measures to combat it – an endeavour that is the focus of Chapter 2.
References
[204] Aalbers, M. (2007), “Place-based and race-based exclusion from mortgage loans: Evidence from three cities in the Netherlands.”, Journal of urban affairs, 29(1), 1-29., https://www.tandfonline.com/doi/full/10.1111/j.1467-9906.2007.00320.x.
[31] Aasebø, T. and I. Willbergh (2022), “Empowering minority students: a study of cultural references in the teaching content.”, Journal of curriculum studies, 54(5), 618-631., https://www.tandfonline.com/doi/full/10.1080/00220272.2022.2095877.
[197] Acolin, A., R. Bostic and G. Painter (2016), “A field study of rental market discrimination across origins in France.”, Journal of Urban Economics, 95, 49-63., https://www.sciencedirect.com/science/article/pii/S0094119016300377.
[150] Adamovic, M. and A. Leibbrandt (2023), “Is there a glass ceiling for ethnic minorities to enter leadership positions? Evidence from a field experiment with over 12,000 job applications.”, The Leadership Quarterly, 34(2), 101655..
[16] Adukia, A. et al. (2023), “What we teach about race and gender: Representation in images and text of children’s books”, The Quarterly Journal of Economics, 138(4), 2225-2285..
[82] Agnew, R. (1992), “Foundation for a general strain theory of crime and delinquency.”, Criminology, 30(1), 47-88..
[205] Aldridge, R. et al. (2018), “Global patterns of mortality in international migrants: a systematic review and meta-analysis.”, The Lancet, 392(10164), 2553-2566.
[43] Alesina, A. et al. (2024), “Revealing Stereotypes: Evidence from Immigrants in schools”, American Economic Review, Vol. 114/7, pp. 1916-1948, https://doi.org/10.1257/aer.20191184.
[123] Alesina, A. and E. La Ferrara (2014), “A test of racial bias in capital sentencing.”, American Economic Review, 104(11), 3397-3433..
[223] Alhusen, J. et al. (2016), “Racial discrimination and adverse birth outcomes: an integrative review.”, Journal of midwifery & women’s health, 61(6), 707-720., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5206968/.
[32] Alrababa’h, A. et al. (2021), “Can exposure to celebrities reduce prejudice? The effect of Mohamed Salah on Islamophobic behaviors and attitudes.”, American Political Science Review, 115(4), 1111-1128., https://www.cambridge.org/core/journals/american-political-science-review/article/can-exposure-to-celebrities-reduce-prejudice-the-effect-of-mohamed-salah-on-islamophobic-behaviors-and-attitudes/A1DA34F9F5BCE905850AC8FBAC78BE58.
[244] Alsan, M., O. Garrick and G. Graziani (2019), “Does diversity matter for health? Experimental evidence from Oakland.”, American Economic Review, 109(12), 4071-4111..
[228] Alzheimer’s Association (2024), “2024 Alzheimer’s Disease Facts and Figures”, https://www.alz.org/media/Documents/alzheimers-facts-and-figures.pdf.
[78] Ammermueller, A. (2012), “Violence in European schools: A widespread phenomenon that matters for educational production.”, Labour Economics, 19(6), 908-922..
[196] Andrés Durà, R. (2021), “Hongares, la inmobiliaria que solo alquila a personas migrantes en València”, La Vanguardia, https://www.lavanguardia.com/local/valencia/20211017/7791700/hongares-inmobiliaria-alquila-personas-migrantes-valencia.html.
[8] Anne, D. et al. (2024), “Discriminations à l’embauche des Ultramarins : un testing en outre-mer et dans l’Hexagone.”, Revue économique, Vol. 75(3), 483-500., https://doi.org/10.3917/reco.753.0483.
[24] Aping et al. (2024), AEWT– Schulbuchanalyse, Advancing Equality Within the Austrian School System, https://lernplattform-aewtass.org/schulbuchanalyse/.
[75] Appel, M. (2012), “Anti‐immigrant propaganda by radical right parties and the intellectual performance of adolescents.”, Political Psychology, 33(4), 483-493..
[74] Appel, M., S. Weber and N. Kronberger (2015), “The influence of stereotype threat on immigrants: Review and meta-analysis.”, Frontiers in Psychology, 6, 147409..
[25] Araújo, M. and S. Maeso (2012), “History textbooks, racism and the critique of Eurocentrism: beyond rectification or compensation.”, Ethnic and Racial Studies, 35(7), 1266-1286., https://www.tandfonline.com/doi/full/10.1080/01419870.2011.600767.
[118] Arnold, D., W. Dobbie and P. Hull (2022), “Measuring Racial Discrimination in Bail Decisions”, American Economic Review, Vol. 112/9, pp. 2992-3038, https://doi.org/10.1257/aer.20201653.
[124] Arnold, D., W. Dobbie and C. Yang (2018), “Racial Bias in Bail Decisions,”, The Quarterly Journal of Economics, Volume 133, Issue 4, November 2018, Pages 1885–1932, https://doi.org/10.1093/qje/qjy012.
[176] Auer, D. (2022), “Firing discrimination: Selective labor market responses of firms during the COVID-19 economic crisis.”, PloS one, 17(1), e0262337., https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0262337.
[97] Auer, T. et al. (2022), “Same Different but not Same Same: Ethnic Discrimination in Application Procedures in the German Public Sector.”, Empirical Economics Letters, 21(7)..
[192] Auspurg, K., A. Schneck and T. Hinz (2019), “Closed doors everywhere? A meta-analysis of field experiments on ethnic discrimination in rental housing markets.”, Journal of Ethnic and Migration Studies, 45(1), 95-114., https://www.tandfonline.com/doi/full/10.1080/1369183X.2018.1489223.
[39] Avvisati, F. et al. (2014), “Getting parents involved: A field experiment in deprived schools.”, Review of Economic Studies, 81(1), 57-83., https://academic.oup.com/restud/article/81/1/57/1729174?casa_token=-O7hHs7Zj0sAAAAA:4p2f3Z8NV6AiCs19q5MeJa7dDkM7XXGtqxFsvVTKxzBAom8bXjOIGL7LAQLL36Uit_YkGcfF3Std.
[105] Bacher-Hicks, A., S. Billings and D. Deming (2024), “The school to prison pipeline: Long-run impacts of school suspensions on adult crime”, American Economic Journal: Economic Policy, Vol. 16/4, pp. 165–193, https://doi.org/10.1257/pol.20230052.
[148] Baert, S. et al. (2018), “Does size matter? Hiring discrimination and firm size.”, International Journal of Manpower, 39(4), 550-566..
[90] Baert, S. et al. (2021), “Student internships and employment opportunities after graduation: A field experiment.”, Economics of Education Review, 83, 102141..
[183] Baptista, I. and E. Marlier (2019), “Fighting homelessness and housing exclusion in Europe: A study of national policies.”, European Social Policy Network (ESPN). European Commission., https://op.europa.eu/en/publication-detail/-/publication/2dd1bd61-d834-11e9-9c4e-01aa75ed71a1.
[36] Barger, M. et al. (2019), “The relation between parents’ involvement in children’s schooling and children’s adjustment: A meta-analysis.”, Psychological bulletin, 145(9), 855..
[9] Bartoš, V. et al. (2016), “Attention Discrimination: Theory and Field Experiments with Monitoring Information Acquisition”, American Economic Review 106 (6): 1437–75, https://www.aeaweb.org/articles?id=10.1257/aer.20140571.
[194] Bayer, P. et al. (2017), “Racial and ethnic price differentials in the housing market.”, Journal of Urban Economics, 102, 91-105., https://www.sciencedirect.com/science/article/pii/S009411901730061X?via%3Dihub.
[164] Bellé, N., P. Cantarelli and P. Belardinelli (2017), “Cognitive biases in performance appraisal: Experimental evidence on anchoring and halo effects with public sector managers and employees.”, Review of Public Personnel Administration, 37(3), 275-294., https://doi.org/10.1177/0734371X17704891.
[209] Berger, M. and Z. Sarnyai (2015), ““More than skin deep”: stress neurobiology and mental health consequences of racial discrimination.”, Stress, 18(1), 1-10., https://www.tandfonline.com/doi/full/10.3109/10253890.2014.989204.
[13] Bertrand, M. and E. Duflo (2017), “Chapter 8 - Field experiments on discrimination”, A. Banerjee and E. Duflo (eds), Handbook of Field Experiments, North Holland (Vol 1, pages 309-393)..
[157] Biernat, M. and D. Kobrynowicz (1997), “Gender-and race-based standards of competence: lower minimum standards but higher ability standards for devalued groups”, Journal of personality and social psychology, 72(3), 544..
[163] Bohnet, I. and S. Chilazi (2025), “Make Work Fair”, Iris Bohnet and Siri Chilazi, https://makeworkfair.com/.
[162] Bohnet, I., O. Hauser and A. Kristal (2022), “Can Gender and Race Dynamics in Performance Appraisals Be Disrupted? The Case of Anchoring.”, https://scholar.harvard.edu/files/iris_bohnet/files/performance_appraisals.10.27.21.pdf.
[89] Bolli, T., K. Caves and M. Oswald-Egg (2021), “Valuable experience: how university internships affect graduates’ income.”, Research in Higher Education, 62(8), 1198-1247..
[213] Bor, J. et al. (2018), “Police killings and their spillover effects on the mental health of black Americans: a population-based, quasi-experimental study.”, The Lancet, 392(10144), 302-310., https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)31130-9/fulltext.
[161] Botelho, T. et al. (2025), “Scale dichotomization reduces customer racial discrimination and income inequality”, https://www.nature.com/articles/s41586-025-08599-7#citeas.
[187] Boustan, L. (2010), “Was postwar suburbanization “white flight”? Evidence from the black migration”, The Quarterly Journal of Economics, 125(1), 417-443..
[189] Bråmå, Å. (2006), “’White Flight’? The Production and Reproduction of Immigrant Concentration Areas in Swedish Cities, 1990-2000”, Urban Studies, Vol. 43/7, pp. 1127-1146, https://doi.org/10.1080/00420980500406736.
[182] Bramley, G. et al. (2022), “Homelessness amongst Black and minoritised ethnic communities in the UK: A statistical report on the state of the nation.”, Heriot-Watt University, Edinburgh, https://pure.hw.ac.uk/ws/portalfiles/portal/67022998/Technical_Report_State_of_the_Nation.pdf.
[121] Brandon, A. and M. O’Connell (2017), “Same Crime: Different Punishment? Investigating Sentencing Disparities Between Irish and Non-irish Nationals in the Irish Criminal Justice System”, The British Journal of Criminology, Vol. 58/5, pp. 1127-1146, https://doi.org/10.1093/bjc/azx080.
[42] Burgess, S. and E. Greaves (2013), “Test scores, subjective assessment, and stereotyping of ethnic minorities.”, Journal of Labor Economics, 31(3), 535-576..
[20] Butković, M. and E. Vidović (2021), “Representation of Racial Diversity in Picturebooks in Teacher Education Programs in the Republic of Croatia.”, Journal of literary education, (4), 50-68..
[216] Butler, C. et al. (2002), “Internalized racism, body fat distribution, and abnormal fasting glucose among African-Caribbean women in Dominica, West Indies.”, Journal of the National Medical Association, 94(3), 143., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2594102/.
[149] Cahuc, P. et al. (2019), When Correspondence Studies Fail to Detect Hiring Discrimination, IZA DP No. 12653, https://www.iza.org/publications/dp/12653/when-correspondence-studies-fail-to-detect-hiring-discrimination.
[92] Campero, S. (2023), “Racial disparities in the screening of candidates for software engineering internships.”, Social Science Research, 109, 102773..
[139] Carcillo, S. and M. Valfort (2025 (forthcoming)), Invisible Barriers. Understanding and Overcoming Discrimination in the Workplace, The MIT Press.
[186] Card, D., A. Mas and J. Rothstein (2008), “Tipping and the Dynamics of Segregation<sup>*</sup>”, Quarterly Journal of Economics, Vol. 123/1, pp. 177-218, https://doi.org/10.1162/qjec.2008.123.1.177.
[54] Carlana, M., E. Ferrara and P. Pinotti (2022), “Implicit stereotypes in teachers’ track recommendations.”, AEA Papers and Proceedings (Vol. 112, pp. 409-414). 2014 Broadway, Suite 305, Nashville, TN 37203: American Economic Association..
[53] Carlana, M., E. La Ferrara and P. Pinotti (2022), “Goals and gaps: Educational careers of immigrant children.”, Econometrica, 90(1), 1-29..
[152] Carrington, W. and K. Troske (1998), “Interfirm segregation and the black/white wage gap.”, Journal of Labor Economics, 16(2), 231-260..
[60] Carter, P. (2006), “Straddling boundaries: Identity, culture, and school.”, Sociology of education, 79(4), 304-328..
[61] Carter, P. (2005), “Keepin’it real: School success beyond Black and White.”, Oxford University Press..
[219] Carter, R. et al. (2019), “A meta-analytic review of racial discrimination: Relationships to health and culture.”, Race and Social Problems, 11, 15-32., https://link.springer.com/article/10.1007/s12552-018-9256-y.
[37] Castro, M. et al. (2015), “Parental involvement on student academic achievement: A meta-analysis.”, Educational research review, 14, 33-46..
[222] Chaney, C. et al. (2019), “Systematic review of chronic discrimination and changes in biology during pregnancy among African American women.”, Journal of Racial and Ethnic Health Disparities, 6, 1208-1217., https://link.springer.com/article/10.1007/s40615-019-00622-8.
[55] Chareyron, S., L. Erb and Y. L’Horty (2023), “Assessing Discrimination in Access to Higher Education”, Annals of Economics and Statistics, Vol. 151, pp. 121-145, https://doi.org/10.2307/48744152.
[76] Chateignier, C. et al. (2009), “French-Arab students and verbal intellectual performance: Do they really suffer from a negative intellectual stereotype?.”, European Journal of Psychology of Education, 24, 219-234..
[87] Cockx, B. and M. Picchio (2013), “Scarring effects of remaining unemployed for long-term unemployed school-leavers.”, Journal of the Royal Statistical Society Series A: Statistics in Society, 176(4), 951-980..
[211] Coimbra, B. et al. (2022), “Meta-analysis of the effect of racial discrimination on suicidality.”, SSM-Population Health, 20, 101283., https://www.sciencedirect.com/science/article/pii/S2352827322002622.
[200] Combes, P. et al. (2018), “Neighbor discrimination theory and evidence from the French rental market”, Journal of Urban Economics, 104, 104-123., https://www.sciencedirect.com/science/article/pii/S0094119018300020?via%3Dihub.
[11] Coutts, A. (2023), “Racial bias around the world”, Retrieved April, 1, 2023.
[109] Coviello, D. and N. Persico (2015), “An economic analysis of Black-White disparities in the New York Police Department’s stop-and-frisk program.”, The Journal of Legal Studies, 44(2), 315-360..
[35] Crozier, G. and J. Davies (2007), “Hard to reach parents or hard to reach schools? A discussion of home—school relations, with particular reference to Bangladeshi and Pakistani parents.”, British educational research journal, 33(3), 295-313., https://www.tandfonline.com/doi/full/10.1080/01411920701243578.
[28] Dee, T. (2005), “A teacher like me: Does race, ethnicity, or gender matter?.”, American Economic Review, 95(2), 158-165.
[114] Défenseur des droits (2017), “Enquête sur Enquête sur, Volume 1, Relations police / population: le cas des contrôles d’identité”, https://juridique.defenseurdesdroits.fr/doc_num.php?explnum_id=16064.
[27] Delhommer, S. (2022), “High school role models and minority college achievement.”, Economics of Education Review, 87, 102222..
[217] Dong, T. et al. (2023), “How discrimination gets under the skin: biological determinants of discrimination associated with dysregulation of the brain-gut microbiome system and psychological symptoms.”, Biological psychiatry, 94(3), 203-214., https://www.sciencedirect.com/science/article/pii/S0006322322017036.
[195] Drydakis, N. (2011), “Ethnic discrimination in the Greek housing market”, Journal of Population Economics, 24, 1235-1255., https://www.jstor.org/stable/41488350.
[144] Drydakis, N. et al. (2018), “Age discrimination in the UK labour market. Does race moderate ageism? An experimental investigation.”, Applied economics letters, 25(1), 1-4..
[143] Drydakis, N., A. Paraskevopoulou and V. Bozani (2022), “A field study of age discrimination in the workplace: the importance of gender and race‒pay the gap.”, Employee Relations: The International Journal, 45(2), 304-327..
[2] ECRI (2021), ECRI’s opinion on the concept of “racialisation”, Adopted at European Commision against Racism and Intolerance’s 87th plenary meeting on 8 December 2021, https://rm.coe.int/ecri-opinion-on-the-concept-of-racialisation/1680a4dcc2#:~:text=The%20use%20of%20the%20concept,raising%2C%20education2%20and%20policy%20making.
[5] ECRI (2017), ECRI General Policy Recommendation no. 2: Equality Bodies to Combat Racism and Intolerance at National Level.
[3] ECRI (2017), ECRI General Policy Recommendation no. 7: On National Legislation to Combat Racism and Racial Discrimination.
[128] Elliott, J. and R. Smith (2004), “Race, Gender, and Workplace Power”, American Sociological Review, Vol. 69/3, pp. 365-386, https://doi.org/10.1177/000312240406900303.
[79] Eriksen, T., H. Nielsen and M. Simonsen (2014), “Bullying in elementary school.”, Journal of Human Resources, 49(4), 839-871..
[86] Eriksson, S. and D. Rooth (2014), “Do employers use unemployment as a sorting criterion when hiring? Evidence from a field experiment.”, American economic review, 104(3), 1014-1039..
[6] European Comission (2023), “Discrimination in the European Union”, Special Eurobarometer 535 Discrimination in the EU April-May 2023, https://europa.eu/eurobarometer/surveys/detail/2972.
[4] European Commission (2020), EU Anti-racism Action Plan 2020-2025, https://commission.europa.eu/strategy-and-policy/policies/justice-and-fundamental-rights/combatting-discrimination/racism-and-xenophobia/eu-anti-racism-action-plan-2020-2025_en (accessed on 4 July 2024).
[214] Fani, N. et al. (2022), “Racial discrimination associates with lower cingulate cortex thickness in trauma-exposed black women.”, Neuropsychopharmacology, 47(13), 2230-2237., https://www.nature.com/articles/s41386-022-01445-8.
[94] Farvaque, N. (2009), “Discrimination in access to the internship: from students’ feelings to teachers’ intervention.”, Employment training. Revue française de sciences sociales, (105), 21-36..
[85] Filomena, M. (2023), “Unemployment scarring effects: An overview and meta-analysis of empirical studies”, Italian Economic Journal, 1-60..
[233] FitzGerald, C. and S. Hurst (2017), “Implicit bias in healthcare professionals: a systematic review.”, BMC medical ethics, 18, 1-18., https://link.springer.com/article/10.1186/s12910-017-0179-8.
[122] Fjællegaard Jensen, M. and A. Manning (2025), Background Matters, but Not Whether Parents Are Immigrants: Outcomes of Children Born in Denmark, American Economic Journal: Applied Economics (Forthcoming), https://www.aeaweb.org/articles?id=10.1257/app.20230389&from=f.
[64] Fordham, S. and J. Ogbu (1986), “Black students’ school success: Coping with the “burden of ‘acting white’”, The urban review, 18(3), 176-206..
[227] Forrester, S. et al. (2019), “A framework of minority stress: From physiological manifestations to cognitive outcomes.”, The Gerontologist, 59(6), 1017-1023., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6858824/.
[153] Forth, J., N. Theodoropoulos and A. Bryson (2023), “The role of the workplace in ethnic wage differentials.”, British Journal of Industrial Relations, 61(2), 259-290..
[179] Fougère, D. et al. (2013), “Social housing and location choices of immigrants in France.”, International Journal of Manpower, 34(1), 56-69., https://www.emerald.com/insight/content/doi/10.1108/01437721311319656/full/html.
[107] FRA (2024), “Addressing Racism in Policing”, European Union Agency for Fundamental Rights, https://fra.europa.eu/sites/default/files/fra_uploads/fra-2024-addressing-racism-in-policing_en.pdf.
[14] FRA (2023), “Being black in the EU ― experiences of people of African descent”, Publications Office of the European Union, https://fra.europa.eu/en/publication/2023/being-black-eu.
[58] Fryer Jr, R. and P. Torelli (2010), “An empirical analysis of ‘acting white’.”, Journal of Public Economics, 94(5-6), 380-396..
[242] Fumarco, L. et al. (2024), “Gender Identity-, Race-, and Ethnicity-Based Discrimination in Access to Mental Health Care: Evidence from an Audit Correspondence Field Experiment.”, American Journal of Health Economics, 10(2), 182-214..
[178] García Martín, A. and E. Buch Sánchez (2020), “¿Se alquila? Racismo y xenofobia en el mercado del alquiler”, Provivienda, Madrid, https://www.provivienda.org/wp-content/uploads/Se-alquila.-Racismo-y-xenofobia-en-el-mercado-del-alquiler.pdf.
[171] Gasser, M., N. Flint and R. Tan (2000), “Reward expectations: The influence of race, gender and type of job.”, Journal of Business and Psychology, 15, 321-329., https://link.springer.com/article/10.1023/A:1007876103048.
[221] Gee, G., K. Walsemann and E. Brondolo (2012), “A life course perspective on how racism may be related to health inequities.”, American journal of public health, 102(5), 967-974., https://pubmed.ncbi.nlm.nih.gov/22420802/.
[246] Geller, J. (1988), “Racial bias in the evaluation of patients for psychotherapy.”, In L. Comas-Díaz & E. E. H. Griffith (Eds.), Clinical guidelines in cross-cultural mental health (pp. 112–134). John Wiley & Sons., https://psycnet.apa.org/record/1988-97772-005.
[198] Ghekiere, A. et al. (2022), “Introducing a vignette experiment to study mechanisms of ethnic discrimination on the housing market.”, PloS one, 17(10), e0276698., https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0276698.
[17] Ghelam, S. (2024), “Where are the characters of non-white children in children’s literature?”, General Ledger, February 22, 2024, https://www.amazon.fr/personnages-denfants-blancs-litt%C3%A9rature-jeunesse/dp/2494060079.
[18] Ghelam, S. (2021), Représentation des personnages enfantins non blancs dans les albums jeunesse publiés en France entre 2010 et 2020, EHESS – Université Paris Nanterre, https://genreed.hypotheses.org/files/2021/08/SGHELAM-M2_compressed.pdf.
[174] Giuliano, L., D. Levine and J. Leonard (2011), “Racial bias in the manager-employee relationship: An analysis of quits, dismissals, and promotions at a large retail firm.”, Journal of human resources, 46(1), 26-52..
[245] Glock, S. (2016), “Does ethnicity matter? The impact of stereotypical expectations on in-service teachers’ judgments of students.”, Social Psychology of Education, 19, 493-509..
[165] Glover, D., A. Pallais and W. Pariente (2017), “Discrimination as a self-fulfilling prophecy: Evidence from French grocery stores”, The Quarterly Journal of Economics, 132(3), 1219-1260., https://doi.org/10.1093/qje/qjx006.
[115] Goncalves, F. and S. Mello (2021), “A few bad apples? Racial bias in policing.”, American Economic Review, 111(5), 1406-1441..
[240] Graham, G. (2016), “Racial and ethnic differences in acute coronary syndrome and myocardial infarction within the United States: from demographics to outcomes”, Clinical cardiology, Vol. 39/5, pp. 299-306, https://doi.org/10.1002/clc.22524.
[108] Gramlich, J. (2019), “From police to parole, black and white Americans differ widely in their views of criminal justice system”, Pew Research Center, https://www.pewresearch.org/short-reads/2019/05/21/from-police-to-parole-black-and-white-americans-differ-widely-in-their-views-of-criminal-justice-system/.
[10] Greenwald, A., D. McGhee and J. Schwartz (1998), “Measuring individual differences in implicit cognition: The implicit association test.”, Journal of Personality and Social Psychology, Vol. 74/6, pp. 1464-1480, https://doi.org/10.1037//0022-3514.74.6.1464.
[175] Grotti, R., S. Aradhya and J. Härkönen (2023), “Hiring or firing? The dynamics of unemployment inequality among the children of immigrants.”, Stockholm Research Reports in Demography, https://su.figshare.com/articles/preprint/Hiring_or_firing_The_dynamics_of_unemployment_inequality_among_the_children_of_immigrants/21915228.
[193] Gusciute, E., P. Mühlau and R. Layte (2022), “Discrimination in the rental housing market: a field experiment in Ireland”, Journal of Ethnic and Migration Studies, 48(3), 613-634., https://www.tandfonline.com/doi/full/10.1080/1369183X.2020.1813017.
[68] Guyon, N. and E. Huillery (2021), “Biased aspirations and social inequality at school: Evidence from french teenagers.”, The Economic Journal, 131(634), 745-796..
[56] Guyon, N., E. Maurin and S. McNally (2012), “The effect of tracking students by ability into different schools: A natural experiment.”, Journal of Human resources, 47(3), 684-721..
[243] Halla, M., C. Kah and R. Sausgruber (2021), “Testing for Ethnic Discrimination in Outpatient Health Care: Evidence from a Field Experiment in Germany”, IZA DP No. 14886, https://www.iza.org/publications/dp/14886/testing-for-ethnic-discrimination-in-outpatient-health-care-evidence-from-a-field-experiment-in-germany.
[234] Hall, W. et al. (2015), “Implicit racial/ethnic bias among health care professionals and its influence on health care outcomes: a systematic review.”, American journal of public health, 105(12), e60-e76., https://ajph.aphapublications.org/doi/abs/10.2105/ajph.2015.302903.
[140] Hangartner, D., D. Kopp and M. Siegenthaler (2021), “Monitoring hiring discrimination through online recruitment platforms.”, Nature, 589(7843), 572-576..
[201] Hanson, A. et al. (2016), “Discrimination in mortgage lending: Evidence from a correspondence experiment.”, Journal of Urban Economics, 92, 48-65., https://www.sciencedirect.com/science/article/pii/S0094119015000868?via%3Dihub.
[231] Hanssens, L. et al. (2017), “Perceived discrimination in primary healthcare in Europe: evidence from the cross‐sectional QUALICOPC study.”, Health & social care in the community, 25(2), 641-651., https://onlinelibrary.wiley.com/doi/10.1111/hsc.12353.
[173] Hargis, M. et al. (2006), “Race differences in termination at work: The role of educational inequality.”, Journal of Business and Psychology, 20, 587-598., https://www.jstor.org/stable/25092959.
[49] Heckman, J. and G. Karapakula (2019), “The Perry Preschoolers at late midlife: A study in design-specific inference”, National Bureau of Economic Research..
[95] Helland, H. and L. Støren (2006), “Vocational education and the allocation of apprenticeships: equal chances for applicants regardless of immigrant background?”, European sociological review, 22(3), 339-351..
[170] Hernandez, M. et al. (2019), “Bargaining while Black: The role of race in salary negotiations”, Journal of Applied Psychology, 104(4), 581, https://www.apa.org/pubs/journals/releases/apl-apl0000363.pdf.
[44] Hinnerich, B., E. Höglin and M. Johannesson (2015), “Discrimination against students with foreign backgrounds: Evidence from grading in Swedish public high schools.”, Education Economics, 23(6), 660-676..
[99] Hirschi, T. and M. Gottfredson (1983), “Age and the explanation of crime”, American journal of sociology, 89(3), 552-584..
[235] Hoffman, K. et al. (2016), “Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites”, Proceedings of the National Academy of Sciences, 113(16), 4296-4301., https://www.pnas.org/doi/full/10.1073/pnas.1516047113.
[135] Hoffman, M., L. Kahn and D. Li (2018), “Discretion in hiring”, The Quarterly Journal of Economics, 133(2), 765-800..
[113] Home Office (2024), “Police powers and procedures: Stop and search and arrests, England and Wales, year ending 31 March 2023 (second edition)”, Table 2.6., https://www.gov.uk/government/statistics/stop-and-search-and-arrests-year-ending-march-2023/police-powers-and-procedures-stop-and-search-and-arrests-england-and-wales-year-ending-31-march-2023#s2.6.
[210] Hudson, D. et al. (2016), “Racial discrimination, john henryism, and depression among African Americans.”, Journal of Black psychology, 42(3), 221-243., https://journals.sagepub.com/doi/pdf/10.1177/0095798414567757.
[177] International Business Times (2016), “’No Blacks’: French estate agency barred black families from renting a flat near Paris”, https://www.ibtimes.co.uk/no-blacks-french-estate-agency-discriminated-black-families-renting-flat-near-paris-1598381.
[41] Jacoby-Senghor, D., S. Sinclair and J. Shelton (2016), “A lesson in bias: The relationship between implicit racial bias and performance in pedagogical contexts.”, Journal of Experimental Social Psychology, 63, 50-55..
[110] Jobard, F. and R. Lévy (2009), “Profiling minorities a study of stop-and-search practices in Paris”, (Doctoral dissertation, Open Justice Initiative)..
[130] Jowell, R. and P. Prescott-Clarke (1970), “Racial Discrimination and White-collar Workers in Britain”, Race, 11(4), 397-417., https://doi.org/10.1177/030639687001100401.
[96] Kaas, L. and C. Manger (2012), “Ethnic discrimination in Germany’s labour market: A field experiment.”, German economic review, 13(1), 1-20..
[129] Kalev, A. (2014), “How You Downsize Is Who You Downsize”, American Sociological Review, Vol. 79/1, pp. 109-135, https://doi.org/10.1177/0003122413518553.
[225] Kennedy-Moulton, K. et al. (2025), “Maternal and infant health inequality: new evidence from linked administrative data (No. w30693).”, Review of Economics and Statistics, Revise and Resubmit. Also available as NBER Working Paper No. 30693, IZA Discussion Paper No. 15745., https://web.stanford.edu/~perssonp/MaternalInfantHealthInequality.pdf.
[93] Kergoat, P. and E. Sulzer (2017), “Mesure et analyse des discriminations d’accès à l’apprentissage.”, Rapport d’évaluation FEJ/INJEP..
[241] Kugelmass, H. (2016), ““Sorry, I’m Not Accepting New Patients” an audit study of access to mental health care”, Journal of Health and Social Behavior, Vol. 57/2, pp. 168-183, https://doi.org/10.1177/0022146516647098.
[112] Lammy, D. (2017), “The Lammy Review: An independent review into the treatment of, and outcomes for, Black, Asian and Minority Ethnic individuals in the Criminal Justice System.”, London: Lammy Review..
[15] Larrick, N. (1965), “The all-white world of children’s books”, Saturday Review, Vol. Sept 11, pp. 63-65, https://brichislitspot.wordpress.com/wp-content/uploads/2017/08/384larrick.pdf.
[166] Lehmann, J. (2011), “Job assignment and promotion under statistical discrimination: Evidence from the early careers of lawyers.”, MPRA Paper University Library of Munich, https://EconPapers.repec.org/RePEc:pra:mprapa:33466.
[145] L’Horty, Y. et al. (2022), “Is disability more discriminatory in hiring than ethnicity, address or gender? Evidence from a multi-criteria correspondence experiment.”, Social science & medicine, 303, 114990..
[138] Li, D., L. Raymond and P. Bergman (2025), “Hiring as exploration”, Review of Economic Studies (Forthcoming), https://danielle-li.github.io/assets/docs/HiringAsExploration.pdf.
[7] Liebig, T. and G. Spielvogel (2021), “Residential segregation of immigrants: Patterns, drivers, effects and policy responses”, in International Migration Outlook 2021, OECD Publishing, Paris, https://www.oecd.org/en/publications/international-migration-outlook-2021_29f23e9d-en/full-report/component-9.html#chapter-d1e51854.
[154] Linos, E., S. Mobasseri and N. Roussille (2025), “Asymmetric Peer Effects at Work: The Effect of White Coworkers on Black Women’s Careers.”, Management Science (Forthcoming), https://ninaroussille.github.io/files/LMR_2023.pdf.
[151] Lippens, L. et al. (2022), “Is labour market discrimination against ethnic minorities better explained by taste or statistics? A systematic review of the empirical evidence.”, Journal of Ethnic and Migration Studies, 48(17), 4243-4276..
[131] Lippens, L., S. Vermeiren and S. Baert (2023), “The state of hiring discrimination: A meta-analysis of (almost) all recent correspondence experiments.”, European Economic Review, 151, 104315..
[62] Lorenz, G., Z. Boda and Z. Salikutluk (2021), “Oppositional culture revisited. Friendship dynamics and the creation of social capital among Turkish minority adolescents in Germany.”, Journal of Ethnic and Migration Studies, 47(17), 3986-4005..
[249] Lorion, R. (1974), “Patient and therapist variables in the treatment of low-income patients.”, Psychological Bulletin, 81(6), 344., https://doi.org/10.1037/h0036484.
[237] Louie, P. and R. Wilkes (2018), “Representations of race and skin tone in medical textbook imagery.”, Social Science & Medicine, 202, 38-42..
[159] Mackenzie, L., J. Wehner and S. Correll (2019), “Why most performance evaluations are biased, and how to fix them.”, Harvard Business Review., https://hbr.org/2019/01/why-most-performance-evaluations-are-biased-and-how-to-fix-them.
[232] Maina, I. et al. (2018), “A decade of studying implicit racial/ethnic bias in healthcare providers using the implicit association test.”, Social science & medicine, 199, 219-229..
[40] Malouff, J. and E. Thorsteinsson (2016), “Bias in grading: A meta-analysis of experimental research findings”, Australian Journal of Education, 60(3), 245-256..
[29] Marx, D. and J. Roman (2022), “Female role models: Protecting women’s math test performance.”, Personality and Social Psychology Bulletin, 28(9), 1183-1193..
[106] McIntyre, N., N. Parveen and T. Thomas (2021), “Exclusion rates five times higher for black Caribbean pupils in parts of England”, The Guardian, https://www.theguardian.com/education/2021/mar/24/exclusion-rates-black-caribbean-pupils-england.
[156] McKay, P. and M. McDaniel (2006), “A reexamination of black-white mean differences in work performance: More data, more moderators.”, Journal of Applied Psychology, 91(3), 538..
[126] McPherson, M., L. Smith-Lovin and J. Cook (2001), “Birds of a Feather: Homophily in Social Networks”, Annual Review of Sociology, Vol. 27/1, pp. 415-444, https://doi.org/10.1146/annurev.soc.27.1.415.
[12] Meissner, F. et al. (2019), “Predicting Behavior With Implicit Measures: Disillusioning Findings, Reasonable Explanations, and Sophisticated Solutions”, Frontiers in Psychology, Vol. 10, https://doi.org/10.3389/fpsyg.2019.02483.
[248] Meyerhoefer, C., B. Xue and A. Poznańska (2025), “Implications of the Decline in LGBT Rights for Population Mental Health: Evidence from Polish “LGBT-free zones””, Journal of Health Economics, https://www.sciencedirect.com/science/article/pii/S0167629625000074.
[22] Mikander, P. (2023), “Racism in Finnish school textbooks: Developments and discussions.”, Finland’s Famous Education System: Unvarnished Insights into Finnish Schooling (pp. 289-302). Singapore: Springer Nature Singapore., https://link.springer.com/chapter/10.1007/978-981-19-8241-5_18.
[111] Miller, J. (2010), “Stop and search in England: A reformed tactic or business as usual?”, The British Journal of Criminology, 50(5), 954-974..
[119] Ministry of Justice (2021), “National statistics: Ethnicity and the Criminal Justice System, 2020”, https://www.gov.uk/government/statistics/ethnicity-and-the-criminal-justice-system-statistics-2020/ethnicity-and-the-criminal-justice-system-2020#defendants.
[238] Mukwende, M., P. Tamony and M. Turner (2020), “Mind the Gap: A Handbook of Clinical Signs in Black and Brown Skin”, St George’s, University of London., https://www.blackandbrownskin.co.uk/mindthegap.
[88] Musset, P. (2019), “Improving work-based learning in schools”, OECD Social, Employment and Migration Working Papers, No. 233, OECD Publishing, Paris, https://doi.org/10.1787/918caba5-en.
[26] Nagre, K. (2023), (Mis)educating England: eurocentric narratives in secondary school history textbooks., Race Ethnicity and Education, 28(1), 134–153., https://doi.org/10.1080/13613324.2023.2192945.
[98] National Association of Colleges and Employers (2020), “Racial Disproportionalities Exist in Terms of Intern Representation”, https://www.naceweb.org/diversity-equity-and-inclusion/trends-and-predictions/racial-disproportionalities-exist-in-terms-of-intern-representation/#:~:text=African%2DAmerican%20students%20are%20more,unpaid%20or%20never%20interns%3B%20and.
[66] Odim, C. and P. Carter (2023), ““Acting White” and Oppositional Culture in Education”, Oxford University Press..
[181] OECD (2024), Challenges to measuring homelessness among migrants in OECD and EU countries.
[84] OECD (2022), “Meeting of the Council at Ministerial Level, 9-10 June 2022”, Recommendation of the Council on Creating Better Opportunities for Young People, https://www.oecd.org/mcm/Recommendation-on-Creating-Better-Opportunities-for-Young-People.pdf.
[48] OECD (2021), “Young People with Migrant Parents, Making Integration Work”, OECD Publishing, Paris, https://doi.org/10.1787/6e773bfe-en.
[127] OECD (2020), “All Hands In? Making Diversity Work for All”, OECD Publishing, Paris,, https://doi.org/10.1787/efb14583-en.
[51] OECD (2019), PISA 2018 Results (Volume II): Where All Students Can Succeed, OECD Publishing, Paris, https://doi.org/10.1787/b5fd1b8f-en.
[47] OECD (2018), Engaging Young Children: Lessons from Research about Quality in Early Childhood Education and Care, Starting Strong, OECD Publishing, Paris, https://doi.org/10.1787/9789264085145-en.
[50] OECD (2017), “Catching Up? Intergenerational Mobility and Children of Immigrants”, OECD Publishing, Paris, https://doi.org/10.1787/9789264288041-en.
[1] OECD/European Commission (2023), “Indicators of Immigrant Integration 2023: Settling In”, OECD Publishing, Paris, https://doi.org/10.1787/1d5020a6-en.
[65] Ogbu, J. (1979), “Minority education and caste: The American system in cross-cultural perspective.”, Crisis, 86(1), 17-21..
[215] Okeke, O. et al. (2023), “Indirect effects of racial discrimination on health outcomes through prefrontal cortical white matter integrity.”, Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 8(7), 741-749., https://www.sciencedirect.com/science/article/pii/S2451902222001203?via%3Dihub.
[83] Ouellet-Morin, I. et al. (2011), “Blunted cortisol responses to stress signal social and behavioral problems among maltreated/bullied 12-year-old children.”, Biological psychiatry, 70(11), 1016-1023..
[34] Parsons, A. et al. (2018), “Parental involvement: Rhetoric of inclusion in an environment of exclusion.”, Journal of Contemporary Ethnography, 47(1), 113-139., https://journals.sagepub.com/doi/full/10.1177/0891241616676874?journalCode=jcec.
[239] Pattillo, M. et al. (2023), “Racism against racialized migrants in healthcare in Europe: a scoping review”, International Journal for Equity in Health, 22(1), 201..
[146] Petit, P., M. Bunel and Y. L’horty (2020), “Discrimination in hiring in the public sphere: the respective effects of address and origin.”, Economic Review, 71(1), 31-56..
[134] Polavieja, J. et al. (2023), “In your face: a comparative field experiment on racial discrimination in Europe.”, Socio-Economic Review, 21(3), 1551-1578..
[100] Quetelet, A. (2003), “Research on the propensity for crime at different ages.”, Crime: Critical Concepts in Sociology; Bean, P., Ed.; Taylor & Francis: London, UK, 119-135..
[133] Quillian, L. et al. (2019), “Do some countries discriminate more than others? Evidence from 97 field experiments of racial discrimination in hiring”, Sociological Science, 6, 467-496..
[142] Quillian, L. and J. Lee (2023), “Trends in racial and ethnic discrimination in hiring in six Western countries.”, Proceedings of the National Academy of Sciences, 120(6), e2212875120..
[191] Quillian, L., J. Lee and B. Honoré (2020), “Racial discrimination in the US housing and mortgage lending markets: a quantitative review of trends, 1976–2016”, Race and Social Problems, 12, 13-28., https://link.springer.com/article/10.1007/s12552-019-09276-x.
[141] Quillian, L., J. Lee and M. Oliver (2020), “Evidence from field experiments in hiring shows substantial additional racial discrimination after the callback.”, Social Forces, 99(2), 732-759..
[132] Quillian, L. et al. (2017), “Meta-analysis of field experiments shows no change in racial discrimination in hiring over time”, Proceedings of the National Academy of Sciences, Vol. 114/41, pp. 10870-10875, https://doi.org/10.1073/pnas.1706255114.
[218] Raney, J. et al. (2023), “Racial discrimination is associated with binge-eating disorder in early adolescents: a cross-sectional analysis.”, Journal of Eating Disorders, 11(1), 139., https://jeatdisord.biomedcentral.com/articles/10.1186/s40337-023-00866-0.
[190] Rathelot, R. and M. Safi (2013), “Local Ethnic Composition and Natives’ and Immigrants’ Geographic Mobility in France, 1982–1999”, American Sociological Review, Vol. 79/1, pp. 43-64, https://doi.org/10.1177/0003122413514750.
[120] Riba, J. et al. (2023), “Equality Before the Courts? Studying Citizenship Disparities in Sentencing in Catalonia.”, European Journal on Criminal Policy and Research, 29(4), 601-624..
[102] Riddle, T. and S. Sinclair (2019), “Racial disparities in school-based disciplinary actions are associated with county-level rates of racial bias.”, Proceedings of the National Academy of Sciences, 116(17), 8255-8260..
[155] Rim, N. et al. (2024), “The black-white recognition gap in award nominations.”, Journal of Labor Economics, 42(1), 1-23..
[230] Rivenbark, J. and M. Ichou (2020), “Discrimination in healthcare as a barrier to care: experiences of socially disadvantaged populations in France from a nationally representative survey.”, BMC public health, 20(1), 31., https://bmcpublichealth.biomedcentral.com/articles/10.1186/s12889-019-8124-z.
[160] Rivera, L. and A. Tilcsik (2019), “Scaling down inequality: Rating scales, gender bias, and the architecture of evaluation.”, American Sociological Review, 84(2), 248-274..
[168] Rodwell, J., R. Kienzle and M. Shadur (1998), “The relationship among work‐related perceptions, employee attitudes, and employee performance: The integral role of communications.”, Human Resource Management: Published in Cooperation with the School of Business Administration, The University of Michigan and in alliance with the Society of Human Resources Management, 37(3‐4), 277-293..
[67] Rosenthal, R. and L. Jacobson (1968), “Pygmalion in the classroom.”, The urban review, 3(1), 16-20..
[202] Ross, S. et al. (2008), “Mortgage lending in Chicago and Los Angeles: A paired testing study of the pre-application process.”, Journal of Urban Economics, 63(3), 902-919., https://www.sciencedirect.com/science/article/pii/S0094119007000903?via%3Dihub.
[169] Roussille, N. (2024), “The role of the ask gap in gender pay inequality.”, The Quarterly Journal of Economics, qjae004., https://academic.oup.com/qje/advance-article/doi/10.1093/qje/qjae004/7608508.
[220] Ruiz-Narváez, E. et al. (2024), “Perceived Experiences of racism in Relation to Genome-Wide DNA Methylation and Epigenetic Aging in the Black Women’s Health Study.”, Journal of Racial and Ethnic Health Disparities, 1-10., https://link.springer.com/article/10.1007/s40615-024-01915-3.
[45] Sahlström, E. and M. Silliman (2024), The Extent and Consequences of Teacher Biases against Immigrants, IZA DP No. 16899, https://www.iza.org/publications/dp/16899/the-extent-and-consequences-of-teacher-biases-against-immigrants.
[77] Sander, A. et al. (2017), “Stereotypenbedrohung als Ursache für geringeren Wortschatzzuwachs bei Grundschulkindern mit Migrationshintergrund.”, Zeitschrift für Erziehungswissenschaft, 1(21), 177-197..
[229] Sanofi (2024), “How Can We Rebuild Trust in Healthcare for Improved Health Outcomes?”, https://www.sanofi.com/en/magazine/social-impact/how-can-we-rebuild-trust-in-healthcare-for-improved-health-outcomes.
[136] Schellmann, H. (2024), “The Algorithm: How AI Can Hijack Your Career and Steal Your Future”, HURST, https://www.hurstpublishers.com/book/the-algorithm/.
[70] Schmader, T., M. Johns and C. Forbes (2008), “An integrated process model of stereotype threat effects on performance.”, Psychological review, 115(2), 336..
[57] Schmader, T., B. Major and R. Gramzow (2001), “Coping with ethnic stereotypes in the academic domain: Perceived injustice and psychological disengagement.”, Journal of Social Issues, 57(1), 93-111..
[23] Schuermans, N. (2013), “Geography textbooks and the reproduction of a racist and ethnocentric world view among young people in Flanders.”, Belgeo. Revue belge de géographie, (4)., https://researchportal.vub.be/en/publications/geography-textbooks-and-the-reproduction-of-a-racist-and-ethnocen.
[208] Selvarajah, S. et al. (2022), “Racism, xenophobia, and discrimination: mapping pathways to health outcomes”, The Lancet, 400(10368), 2109-2124., https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(22)02484-9/fulltext.
[224] Sheikh, J. et al. (2022), “Effects of race and ethnicity on perinatal outcomes in high-income and upper-middle-income countries: an individual participant data meta-analysis of 2 198 655 pregnancies.”, The Lancet, 400(10368), 2049-2062., https://pubmed.ncbi.nlm.nih.gov/36502843/.
[188] Shertzer, A. and R. Walsh (2019), “Racial sorting and the emergence of segregation in American cities”, Review of Economics and Statistics, 101(3), 415-427., https://direct.mit.edu/rest/article/101/3/415/58527/Racial-Sorting-and-the-Emergence-of-Segregation-in.
[206] Shor, E. and D. Roelfs (2021), “A global meta-analysis of the immigrant mortality advantage.”, International Migration Review, 55(4), 999-1028..
[167] Siebers, H. and J. Van Gastel (2015), “Why migrants earn less: In search of the factors producing the ethno-migrant pay gap in a Dutch public organization”, Work, employment and society, Vol. 29/3, pp. 371-391, https://doi.org/10.1177/0950017014568138.
[103] Skiba, R. et al. (2011), “Race is not neutral: A national investigation of African American and Latino disproportionality in school discipline. S”, School psychology review, 40(1), 85-107..
[104] Skiba, R. et al. (2002), “The color of discipline: Sources of racial and gender disproportionality in school punishment.”, The urban review, 34, 317-342..
[184] Skifter Andersen, H. et al. (2016), “The impact of housing policies and housing markets on ethnic spatial segregation: Comparing the capital cities of four Nordic welfare states”, International Journal of Housing Policy, 16(1), 1-30., https://www.tandfonline.com/doi/full/10.1080/14616718.2015.1110375.
[125] Small, M. and D. Pager (2020), “Sociological perspectives on racial discrimination.”, Journal of Economic Perspectives, 34(2), 49-67..
[117] Søndergaard, J. and T. Hussein (2022), “ETHNIC PROFILING: MAIN RESULTS FROM THREE STUDIES”, The Danish Institute for Human Rights, https://menneskeret.dk/files/media/document/Ethnic%20profiling%20-%20main%20results%20from%20three%20studies%2C%20April%202022.pdf.
[226] Spears Brown, C. and R. Bigler (2005), “Children’s perceptions of discrimination: A developmental model.”, Child development, 76(3), 533-553., https://srcd.onlinelibrary.wiley.com/doi/10.1111/j.1467-8624.2005.00862.x.
[73] Spencer, S., C. Logel and P. Davies (2016), “Stereotype threat.”, Annual review of psychology, 67, 415-437..
[46] Sprietsma, M. (2013), “Discrimination in grading: Experimental evidence from primary school teachers.”, Empirical economics, 45, 523-538..
[33] Starck, J. et al. (2020), “Teachers are people too: Examining the racial bias of teachers compared to other American adults.”, Educational Researcher, 49(4), 273-284., https://journals.sagepub.com/doi/pdf/10.3102/0013189X20912758?casa_token=Y2RAU9sDmvgAAAAA:ncmagk0D8VVLBg5x9hJ6WfVI_Qgk9nrWqxUUMFewFp4myKTmlU4mQ1YGQz3t-GgDNYLTjwAJsxkKpQ.
[69] Steele, C. (1997), “A threat in the air: How stereotypes shape intellectual identity and performance.”, American psychologist, 52(6), 613..
[72] Steele, C. and J. Aronson (1995), “Stereotype threat and the intellectual test performance of African Americans”, Journal of personality and social psychology, 69(5), 797..
[203] Stefan, M. et al. (2018), “Ethnical discrimination in Europe: Field evidence from the finance industry.”, PloS one, 13(1), e0191959., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5788365/.
[116] Stelter, M. et al. (2022), “Racial bias in police traffic stops: White residents’ county-level prejudice and stereotypes are related to disproportionate stopping of Black drivers.”, Psychological science, 33(4), 483-496..
[91] Sterling, A. and R. Fernandez (2018), “Once in the door: Gender, tryouts, and the initial salaries of managers.”, Management Science, 64(11), 5444-5460..
[59] Stinson, D. (2011), “When the “burden of acting White” is not a burden: School success and African American male students”, The Urban Review, 43, 43-65..
[71] Stone, J. et al. (1999), “Stereotype threat effects on Black and White athletic performance”, Journal of personality and social psychology, 77(6), 1213..
[247] Teasdale, A. and C. Hill (2006), “Preferences of therapists-in-training for client characteristics.”, Psychotherapy: Theory, Research, Practice, Training, 43(1), 111., https://doi.org/10.1037/0033-3204.43.1.111.
[21] Terra, L. and P. Bromley (2012), “The globalization of multicultural education in social science textbooks: Cross-national analyses, 1950–2010.”, Multicultural Perspectives, 14(3), 136-143., https://www.tandfonline.com/doi/full/10.1080/15210960.2012.696980.
[19] Thiery, N. and V. Francis (2015), Figures et représentations de l’enfant noir dans les albums pour la jeunesse, Spirale - Revue de recherches en éducation Année 2015 Suppl.E 55 pp. 39-57, https://www.persee.fr/doc/spira_2118-724x_2015_sup_55_1_1736.
[101] U.S. Department of Education (2023), “Student Discipline and School Climate in U.S. Public Schools”, 2020-21 Civil Rights Data Collection.
[52] Van Ewijk, R. (2011), “Same work, lower grade? Student ethnicity and teachers’ subjective assessments.”, Economics of Education Review, 30(5), 1045-1058..
[63] Van Tubergen, F. and M. van Gaans (2016), “Is there an oppositional culture among immigrant adolescents in the Netherlands?”, Youth & Society, 48(2), 202-219..
[180] Verdugo, G. (2011), “Public housing and residential segregation of immigrants in France, 1968-1999”, Population, 66(1), 169-193, https://www.jstor.org/stable/41337161.
[185] Verdugo, G. and S. Toma (2018), “Can public housing decrease segregation? Lessons and challenges from non-European immigration in France: Can public housing decrease segregation?”, Demography, 55(5), 1803-1828., https://link.springer.com/article/10.1007/s13524-018-0705-4.
[147] Villadsen, A. and J. Wulff (2018), “s the public sector a fairer employer? Ethnic employment discrimination in the public and private sectors.”, Academy of Management Discoveries, 4(4), 429-448..
[81] Vitoroulis, I. and K. Georgiades (2017), Bullying among immigrant and non-immigrant early adolescents: School- and student-level effects, Journal of AdolescenceVolume 61, Issue 1 p. 141-151, https://doi.org/10.1016/j.adolescence.2017.10.008.
[236] von Eisenhart Rothe, Y. (2024), “Diagnose “Morbus Mediterraneus”: Das rassistische Klischee von wehleidigen Migrantinnen”, Spiegel Politik, https://www.spiegel.de/politik/diskriminierung-in-der-medizin-das-rassistische-klischee-von-wehleidigen-migrantinnen-a-7eced19d-851a-406e-aeb8-bea60ae28873?sara_ref=re-xx-cp-sh.
[30] Walker, V. (2001), “African American teaching in the South: 1940–1960.”, American educational research journal, 38(4), 751-779..
[207] Wallace, M., L. Hiam and R. Aldridge (2023), “Elevated mortality among the second-generation (children of migrants) in Europe: what is going wrong? A review”, British Medical Bulletin, 148(1), 5-21., https://academic.oup.com/bmb/article/148/1/5/7335807.
[212] Wallace, S., J. Nazroo and L. Bécares (2016), “Cumulative effect of racial discrimination on the mental health of ethnic minorities in the United Kingdom.”, American journal of public health, 106(7), 1294-1300., https://ajph.aphapublications.org/doi/abs/10.2105/AJPH.2016.303121?journalCode=ajph.
[199] Weichselbaumer, D. and H. Riess (2024), “Discrimination in the Austrian rental housing market: The effect of information concerning first and second-generation immigrant status”, Journal of Housing Economics Volume 66, December 2024, 102030, https://www.sciencedirect.com/science/article/pii/S1051137724000494.
[38] Wilder, S. (2023), “Effects of parental involvement on academic achievement: a meta-synthesis.”, Mapping the Field (pp. 137-157). Routledge..
[158] Williams, J. et al. (2021), “How one company worked to root out bias from performance reviews.”, Harvard Business Review (21 April 2021)., https://hbr.org/2021/04/how-one-company-worked-to-root-out-bias-from-performance-reviews.
[172] Wilson, G. (2005), “Race and job dismissal: African American/White differences in their sources during the early work career.”, American Behavioral Scientist, 48(9), 1182-1199., https://journals.sagepub.com/doi/10.1177/0002764205274815?icid=int.sj-abstract.similar-articles.8.
[80] Wolke, D. et al. (2010), “Bullying and victimization of primary school children in England and Germany: Prevalence and school factors.”, British journal of psychology, 92(4), 673-696..
[137] Yin, L., D. Alba and L. Nicoletti (2024), “OpenAI’s GPT is a recruiter’s dream tool. Test show there’s racial bias”, Bloomberg, https://www.bloomberg.com/graphics/2024-openai-gpt-hiring-racial-discrimination/?utm_source=website&utm_medium=share&utm_campaign=email&leadSource=uverify%20wall.
← 1. Published in the 11 September 1965 issue of The Saturday Review, this study by Librarian Nancy scrutinised 5 206 children’s books released by 63 publishers between 1962 and 1964. Her findings were striking: only 349 books, or 6.4%, depicted one or more Black characters in their illustrations.
← 2. This finding, however, is not consistent across all contexts. For example, experimental evidence from Germany indicates a different pattern of bias: teachers exhibit harsher grading towards below-average racial/ethnic minority students, aligning with their lower expectations, rather than towards above‑average students from these groups. See (Glock, 2016[245]).
← 3. This reduced intensity of screening is primarily because firms incur lower monitoring costs for interns, who are expected to be managed only for a few months, as opposed to full-time hires who are generally expected to stay longer. Furthermore, the implications of any errors in screening are generally less severe for internships or apprenticeships than for full-time positions.
← 4. It is crucial to emphasise that the negative impact of discrimination on mental health has also been confirmed in Europe, although not with a focus on racial/ethnic minorities. Notably, a study focusing on LGBTIQ+ individuals, another group particularly at risk of discrimination, provided the first comprehensive analysis of this issue (Meyerhoefer, Xue and Poznańska, 2025[248]). This study examined the effects of anti-LGBTIQ+ laws implemented by provincial, county, and municipal governments in Poland between 2019 and 2020 on the population’s mental health. Utilising county-level data from 2017 to 2020, the researchers employed difference‑in-differences models to compare changes in suicide attempts, suicides, and overall mortality in areas that enacted anti-LGBTIQ laws with those that did not. The study found that annual suicide attempts increased by 16%, or 5 attempts per 100 000 people, following the enactment of these statutes. Furthermore, the researchers discovered an increase in suicide attempts in areas that considered but ultimately rejected anti-LGBTIQ resolutions, demonstrating that even the mere threat of discrimination against minority groups can lead to declines in mental health.
← 5. Instead, the observed disparities are possibly influenced by bias. For instance, research suggests that therapists favour patients who are “psychologically minded” – those capable of understanding and interpreting human behaviour from a psychological perspective (Teasdale and Hill, 2006[247]). Notably, Black patients are perceived by psychiatrists as being less articulate compared to their White counterparts (Geller, 1988[246]). This perception is potentially compounded by therapists’ negative stereotypes about help seekers of disadvantaged socio‑economic background, who are deemed hostile and untreatable (Lorion, 1974[249]).