Introduction

Discrimination against ethnic minorities constitutes a grave social problem with negative implications for individuals and ethnic groups across the globe (Vines et al. 2017; Kirkinis et al. 2018; Flage 2018; de Freitas et al. 2018; Yip 2018). From prejudice and biases to systemic inequalities and segregation, members of various ethnic groups face unequal treatment across multiple domains of life, including healthcare, education, housing, employment, and sports. Discrimination hinders contact between ethnic groups, leads to further social segregation, distrust in institutions, and detrimentally lowers the overall wellbeing of those affected (Contrada et al. 2001; Smith and Silva 2011; Priest et al. 2013; Cave et al. 2020). This study investigates ethnic discrimination against Roma children in Hungary between the ages of seven and thirteen years in three different team sports: basketball, volleyball, and soccer, also considering the intersection of class and ethnicity.

Some previous studies focused on countries of Western and Northern Europe exploring discrimination in youth sports among migrant communities, but less so on non-immigrant ethnic minority groups. For instance, Flensner and her colleagues (2020) explored how sport club leaders facilitate the integration of immigrants through sport in Sweden. Similarly, Burrman and his colleagues (2017) conducted interviews to further understand how ethnic identity intersects with belonging through the lens of sports, among immigrant youths in Germany. However, in the Central-Eastern European region, where prejudice and discrimination against ethnic minorities is high and the level of migration is relatively low (Gödri, 2015 and 2018), there has been less research on discrimination in sport. Sport could be an ideal tool for vertical social mobility (Földesiné Szabó et al. 2010; Spaaij et al. 2019) and social integration of various groups (Doidge et al. 2020; Flensner et al. 2020; Smith et al. 2019). Sport activities under optimal conditions, and in an inclusive environment, have the potential to reduce prejudice by fostering positive group-dynamics, supporting intergroup contact (Cunningham 2011; Mousa 2020; Lowe 2021) and through de-categorisation and re-categorisation processes (Cunningham 2011). However, unfortunately, it often rather becomes a potential field of discrimination (Földesi, 2010; Vandermeerschen et al. 2015; Kassimeris 2021; Nesseler et al. 2019; Gomez-Gonzalez et al. 2021; Storm et al. 2023; Nesseler et al. 2023). Although, there is limited research in the Central-Eastern European context, previous studies from elsewhere that tested the field of soccer, showed significant discrimination against migrant men (Nesseler et al. 2019; Gomez-Gonzalez et al. 2021; Dur et al. 2023) and women (Storm et al. 2023), against national language minorities (Nesseler et al. 2019), and against Jewish men (Nesseler et al. 2023). However, these findings are limited to soccer and do not automatically predict our context, specifically regarding the intersection of ethnicity and social class. Childhood is a sensitive period for social identity processes, even more so for ethnic minority children, whose experiences of discrimination in childhood may have long-lasting effects (Priest et al. 2013; Cave et al. 2020). Sport participation behaviors among children and youngsters have been mainly researched based on socio-economic status and less so related to ethnicity. Generally, the lack of sport participation was found to be correlated with low socio-economic status (Stuart et al. 2009; Kamphuis et al. 2008; Stalsberg and Pedersen 2010; Cocker et al. 2012).

Stereotypes against members of ethnic minority groups often intersect with those related to people belonging to lower socio-cultural class (Urbiola et al. 2022), especially when members of ethnic minority groups are perceived as distant (Hagendoorn 2010) and particularly in countries with less diversity (Bai et al. 2020). Therefore, it is important to understand how this intersection of ethnicity and class is mirrored in the discrimination of ethnic minority children in team sports.

The current study is inspired by a recent paper on ethnic discrimination in women’s amateur soccer among Scandinavian adults (Storm et al. 2023). We implemented our study in the context of Central-Eastern Europe and expanded the work by Storm and colleagues in the following ways: (a) including different types of sports, (b) testing the discrimination of Roma children, belonging to the most numerous ethnic minority group of Europe, (c) including gender, as a secondary experimental variable beyond ethnicity, and (d) uncovering the intersection between ethnicity and social class. We selected three popular team sports, basketball, volleyball, and soccer for our field experiments, as team sports can provide ideal opportunities for positive contact experiences among children from diverse ethnic backgrounds. These sports are also among the most played and followed globally (Hulteen et al. 2017) while having different gender compositions.

To explore ethnic discrimination, we conducted two online field experiments, following the logic of correspondence studies (Verhaeghe 2020). In Study 1, we set up accounts for mothers approaching basketball and volleyball clubs with non-Roma and Roma-sounding names, inquiring about a try-out practice for their child in a short email. In Study 2, testing the intersectionality between ethnicity and social class, soccer clubs were contacted by the same mother email accounts, having the emails written by Roma accounts coupled with a signifier of low social status.

Background

Roma people in Europe

Roma people are the largest transnational ethnic minority in Europe, with 10–12 million Roma people living in the European Union and the Western Balkans (FRA 2014). Hungary is a relatively homogenous EU country in terms of ethnicity. The largest ethnic minority of Hungary are the Roma people, who make up approximately 6–8% of the total population (Kemény and Béla 2006). In EU member states where Roma people constitute a significant ethnic minority, members of the Roma community frequently experience unequal treatment in several aspects of their daily lives. This is also the case in Hungary, where Roma people face discrimination in the labor market (Bartoš et al. 2016), access to education and health care (FRA 2017), police stop-and-search practices (Miller et al. 2008), and housing (FRA 2016). Furthermore, anti-Roma prejudice is widespread in Hungary (Váradi 2014; Kende et al. 2017; Váradi et al. 2021).

Discrimination against Roma people occurs in many domains of their daily lives, including access to public services, though our scientific knowledge in this field is limited. Most recently, however, correspondence studies conducted in the CEE region (in Hungary and in the Czech Republic) provided empirical evidence on discrimination against Roma people in accessing various kinds of public services (Csomor et al. 2021, Simonovits et al. 2022, Mikula and Montag 2023). A recent regional study (fielded in Hungary, Bulgaria and in the Czech Republic) also found that Roma people are discriminated against while accessing public services in the region. They were found to receive worse treatment in various public services, i.e. school enrolment, access to municipal social housing, or renting municipal premises (Scharle and Erős 2022). Most of these experimental studies agree that ethnicity and low socio-economic status are two intertwined causes of discrimination, concluding that as Roma people are overrepresented among social groups of lower socio-economic status in the examined countries, the two sources of discrimination are cumulative for them (Csomor et al. 2021; Scharle and Erős 2022; Mikula and Montag 2023). Regarding the experiences of Roma children in Hungary, school dropout rates among Roma high school students (48% in 2014) are also significantly higher compared to non-Roma students (9% in 2014) (Hajdu et al. 2014). Therefore, Roma children are not only excluded from afterschool leisure and competitive sport clubs, but also from school sport activities (Földesi 2010).

Research context: sport

Sport is often utilized as a tool for countries’ identity building, for the fostering of a positive international image, and for gaining external political legitimacy (Garamvölgyi and Dóczi 2021). Recent research suggests that international sporting achievements do not substantially influence national pride (Storm and Jakobsen 2020), however, individuals living in less democratic regimes report greater national pride after single instances of international sporting success (Meier and Mutz 2018). Furthermore, the overall influence of the success of national teams on mass participation in sports appears to be overestimated (Haut and Gaum 2018).

Since 2010, the Hungarian government has devoted a considerable amount of political and financial support to the sport sector, especially soccer (Molnar and Whigham 2019). By 2020, a total of 32 soccer stadiums were either built from scratch or renovated, with an estimated budget of 601 million Euros from public funds (Garamvölgyi and Dóczi 2021). The so-called TAO system, which allows corporations to donate up to 70% of their corporate tax to professional sports clubs, was approved by the European Commission in 2011.

With the support of TAO, the total income of the sport associations and federations were 240 billion Hungarian Forints between 2011 and 2017, out of which 70% went to soccer (Gősi 2019). In addition to soccer, the TAO system supports an additional five spectator sports: basketball, handball, water polo, ice hockey and volleyball (Gősi 2019).

Throughout Hungary’s sport history, soccer widely caught the attention of spectators, especially since the 1950s, with the outstanding performance of the ‘Magical Magyars’, which was led by the famous soccer player Ferenc Puskás (Molnar 2007). With its historical successes, and new and accessible open-air stadiums around the cities of Hungary, soccer is considered to be among the most popular Hungarian sports. Traditionally, soccer in Hungary is a male-dominated sport. Basketball, another sport supported by the TAO system, has built its foundation in the country since the 1946 bronze medal win in the EuroBasket Championship, thus becoming another popular sport in Hungary, followed and played by many. Utilizing the TAO system, numerous basketball clubs and academies throughout the country have built new facilities and developed the conditions for the new generation of basketball players, making it one of the most played sports in Hungary. Currently, the Hungarian Basketball Federation supports teams from junior levels (children aged 7–11) to elite first league clubs, both for men and women, including its famous University League (MKOSZ 2023). Regarding volleyball, between 2010 and 2017 the yearly financial income of the Hungarian Sport Federation had increased by almost ten times and in 2017, Volleyball was included into the TAO system leading to further increase in financial growth (Gősi 2019). The increased financial support, similar to basketball, also positively influenced the conditions and resulted in an upturn in the popularity of the sport. Volleyball is most popular among girls and women, with over two thirds of the teams being female teams (Bukta and Gősi 2018).

According to research measuring stereotypical views of sports, whether they are perceived as feminine, masculine or neutral, soccer was found to be considered as a masculine sport, while basketball as gender-neutral and volleyball considered feminine (Metheny 1965; Hardin and Greer, 2009). Research conducted in Hungary confirms these results, as in 2018 the percentage of licensed underage female athletes were 6.4 in soccer, 41.67 in basketball and 67.91 in volleyball (Bukta and Gősi 2018).

When it comes to the role of sport for social inclusion in Hungary, empirical studies indicated that the perception of the impact of sport is positive, however, it does not effectively reach marginalized groups (Török 2018). One of the main reasons being the individuals’ socio-economic background which has also been linked to their gender and ethnic belonging (Földesi 2010). According to various empirical studies conducted in Hungary, the highest proportion of excluded people, including both genders and all ages, belong to the Roma population (Velenczei and Gál 2011, Földesi 2010). The conclusions were derived from in-depth interviews with key figures within the Hungarian sport sector, such as sport leaders, managers and physical education teachers. In addition, they also analyzed sport policies aimed at addressing social exclusion in sports in Hungary (Földesi 2010).

Even though many sporting federations recognize this problem and acknowledge that they lose numerous potential talents due to the economic and cultural capital required for participation, only few of them have taken steps to find a solution. One of the most common examples in Hungary is the Boxing Federation, which collaborates with external sponsors to support individuals from disadvantaged regions (Földesi 2010).

Theories

Organized sports have immense potential to stimulate socio-psychological processes that contribute to social cohesion and the reduction of prejudice (Cunningham 2011; Peachey et al. 2015). While there are still many questions regarding the most effective ways to reduce prejudice (Paluck and Green 2009; Paluck et al. 2021), some interventions have shown promising results in sports (Mousa 2020; Lowe 2021; Peachey et al. 2015). The contact hypothesis is one effective approach for prejudice reduction (Pettigrew and Tropp 2006). It states that four conditions need to be met for contact to reduce prejudice: intergroup cooperation, common goals, equal status, and support by authorities (Allport 1954). These conditions were later redefined as facilitators and intergroup friendship was identified as an ideal form of contact for prejudice reduction (Pettigrew 1998). Team sports are widely regarded as an optimal setting to apply Allport’s contact hypothesis (Brown and Stone 2016; Peachey et al. 2015; Schulenkorf and Sherry 2021), providing opportunities for interaction and collaboration towards common goals (Cotterill 2012). This shared identity fosters a sense of belonging, reducing prejudice (Tuncer 2022).

The social identity framework (Tajfel 1982; Tajfel and Turner 2004) also applies to sports, offering avenues for prejudice reduction (Cunningham 2011; Campo et al. 2019). Team sports can enhance decategorization (Brewer and Miller 1984 in Cunningham 2011) and recategorization (Gaertner et al. 1999 in Cunningham 2011). Decategorization involves learning distinct characteristics of diverse outgroup members and perceiving them as individuals (Brewer and Miller 1984). This process can occur through cooperative games. Recategorization, activated by participation in the same tournament shifts focus to new, common categories (Gaertner et al. 1999).

Sports can optimally reduce prejudice among children in dominant social groups (Brown and Stone 2016; Cunningham 2011; Peachey et al. 2015), and support positive socio-psychological processes, such as promoting self-efficiency (Peachey et al. 2015), self-esteem and self-confidence (Ha and Lyras 2013). These benefits are crucial for ethnic minority children, as an increased self-esteem can mitigate stereotype threat and deidentification (Tajfel 1982; Aronson et al. 1998). For children and adolescents with migrant backgrounds, sports support acculturation through cooperation and competition (Morela et al. 2016). Such positive outcomes, however, are only possible if ethnic minority children and children with lower socio-economic status get the chance to participate in team sports along with children from the majority group.

Literature review

In recent decades, field experiments have measured discrimination in various social domains. Most studies focused on labor market access (Quillian et al. 2019; Zschirnt and Ruedin 2015) access to housing (Flage 2018; Auspurg et al. 2019) and the sharing economy (Abramova 2020). In amateur and recreational sports, perceived economic risk, that was identified as a driver for discrimination in housing and labor, might be less pronounced in the domain of sports (Storm et al. 2023).

Discrimination in sports can hinder participation or lead to unequal treatment during activities. Ethnic and class discrimination in sport has been the focus of several studies (Long et al. 2009; Spaaij et al. 2015). Discrimination was found in all areas, ranging from amateur sports (Higgins and Dale 2013; Adair and Rowe 2010; Burrmann et al. 2017; van Haaften 2019) to elite and professional sports (Cranfield et al. 2016; Meier and Leinwather 2013; Seiberth et al. 2017). Discrimination also relates to gender (Andersson 2011; Elling and Knoppers 2005; Carter-Francique and Flowers 2013; O’Connor 2023) and race (Bradbury et al. 2020; Nauright and Wiggins 2016; Evans et al. 2020).

Discrimination in team sports can occur at various stages of joining a team and participating in activities (Kilvington and Price, 2017). Literature often focuses on discrimination at an advanced stage, such as a professional player being excluded due to ethnicity (Kassimeris, 2021; Földesiné Szabó, 2010). However, for example, children from disadvantaged backgrounds may not even register for clubs due to sport-related costs (Földesi, 2010). Upon acceptance, marginalized players may face unequal treatment from teammates, coaches, staff and spectators (Bennett and Jönsson, 2017), manifesting in derogatory language, limited playtime, and restricted advancement opportunities (Kanazawa and Funk, 2001). They might also be excluded at the entry point, removing their opportunity to play sports (Nesseler et al. 2019; Gomez-Gonzalez et al. 2021; Storm et al. 2023; Nesseler et al. 2023), which is the focus of the present study.

Individual-level discrimination is often rooted in stereotyping. Negative stereotypes about the outgroup justify discrimination and substitute accurate information with generalized, negative images (Dovidio et al. 2010). When stereotypes are activated, outgroup membership becomes more prevalent, enabling discriminatory decisions (Kunda and Spencer 2003; Bodenhausen and Richeson, 2010). Stereotypes about Roma people in Hungary are overwhelmingly negative, depicting them as non-competent, non-trustworthy, and prone to criminality (Kende et al. 2021). Specifically, stereotypes about Roma families being poor are coupled with the notion of poverty being their fault (Örkény 2007; Bernáth and Messing 2013). One stereotype related to perceived poverty is that Roma families have numerous children solely for family benefits (Schwarcz 2012). A recent study identified negative stereotypes as a main driver of discrimination against Roma people in the Hungarian rental housing market (Váradi et al. 2023).

Beyond poverty and ethnicity, intersections of gender and ethnicity mobilize different stereotypes and discriminatory dynamics (Caudwell 2017). Gendered stereotypes towards Roma men and women might play a role in their discrimination in sports, however research in this topic is lacking. In the UK, gender disparity in sports participation is greater among ethnic minorities, and studies often focus on men’s experiences in sports (Long et al. 2009).

Overview of studies

The present studies aim to fill the gap in literature on discrimination in sports against Roma people. We built upon the work conducted by Storm and their colleagues studying discrimination in women’s soccer in Scandinavia in the following ways: by (a) including more types of sports besides soccer, (b) assessing discrimination against Roma children, who constitute the largest ethnic minority group in Europe, (c) incorporating gender as secondary experimental variable alongside ethnicity, and (d) exploring the intersectionality of ethnicity and socioeconomic status. We focused on basketball, volleyball, and soccer, for our field experiments recognizing that these sports are the most popular team sports globally (Hulteen et al. 2017; Ipsos 2021). By taking into account the success and historical impact these sports have had in Hungary, including investments through facilities, academies and talent programs, these team sports are among the six most supported and widely played so-called spectator team sports. The specific selection of these three sports from the six is based on their distinct gender compositions (Bukta and Gősi 2018) enabling us to analyze gender as secondary experimental variable.

To measure ethnic discrimination, we conducted two online field experiments, employing the methodology of correspondence testing. According to Verhaeghe (2020), correspondence testing is the “golden standard” to examine discrimination in ecologically valid settings. We tested discrimination by setting up mock email accounts reaching out to basketball, volleyball, and soccer clubs. Two consecutive studies were conducted to measure ethnic discrimination against Roma children (Study 1) and its intersection with social class (Study 2).

Study 1

The aim of Study 1 was to test whether there is discrimination against Roma children applying to a try-out practice in basketball in the age group of U11 and in volleyball in the group of U13, which is the starting age for both sports.

Methods and materials

We created mock email accounts for mothers reaching out to basketball and volleyball clubs using non-Roma (control group) and Roma-sounding (treatment group) female names, inquiring in brief email messages about try-out opportunities for their children. Since the number of available sport clubs in both sports was less than 100 cases in each field, a within-subject (paired) design was applied.Footnote 1 Based on the typical gender representation in basketball and volleyball, half of the basketball clubs were approached by mothers with sons, and half of them by mothers with daughters, whereas for volleyball, only mothers with daughters inquired. Based on oral consultations with experts in the preparation phase of the study we decided to create exclusively mothers’ profiles, as stereotypically mothers are the ones who reach out to coaches. Our experiment was preregistered on osf.io.Footnote 2

Participants

We tested the full population of sport clubs (i.e. all clubs in the given sport and the given age group) in the two sports (volleyball and basketball). We included all volleyball clubs offering trainings and competitions for girls aged under 13 (N = 61). Their email addresses were collected from the official website of the Hungarian Volleyball FederationFootnote 3. All basketball clubs that train children under the age of 11 (N = 76) were identified through the official website of the Hungarian Basketball FederationFootnote 4 and included in our study. We approached altogether 137 sport clubs. While we aimed to identify the correct email addresses, in a few cases there was an error in the mail delivery, so those observations were excluded from the analysis. We tested discrimination in the youngest age group for both sports, in order to avoid biases in responses based on assumed previous experience of the players. The final sample sizes and other details of the fieldwork can be found in Table S1 in the Supplementary materials.

Procedure

Following the research design of Gomez-Gonzalez et al. (2021) and Storm et al. (2023) we approached each sport club with the following two types of e-mail texts (translated into Hungarian). Emails were sent during registration season in September 2023. The two types of texts were constructed in a way that their content would be identical, however the wording would differ to avoid bias in responses. The two texts were randomly assigned to the Roma and non-Roma profiles and were sent to each club one day apart.Footnote 5

Version 1:

Subject: Applying for training

Dear Coach!

My name is [mother’s full name] and my daughter/son, [child’s first name], was born in 2011/2013 and we would like for him/her to take volleyball/basketball more seriously.

I was wondering if it would be possible for him/her to attend a training session at your club.

Thank you in advance for your reply!

Best regards:

[mother’s full name]

Version 2:

Subject: Interest in training

Dear Association,

I am [mother’s full name], my daughter/son [child’s first name] is now 12/10 years old and would like to start playing volleyball/basketball. I would like to know if she/he could join the training sessions of the association?

Have a nice day, [mother’s full name]

Responses and non-responses to the emails were recorded within the period of 11 days. In case of any response, a polite email informing the club that the child was no longer interested in participating in the practice for practical reasons was sent within 8 days. All participants of the research were debriefed within five months (see Appendix 2).

Measures

The primary experimental variable was ethnicity, which was indicated by typically Roma and non-Roma sounding names of the mothers and their children. The secondary experimental variable was gender which we only used in case of basketball according to its mixed gender representation.

In line with previous experimental studies in the fields of sports by Gomez-Gonzalez et al. (2021) and Storm et al. (2023), we created four typically Roma sounding and four non-Roma sounding names for both genders. The proper selection of names is crucial in these kinds of online experiments, especially because the names served as the only experimental stimuli, since we did not provide any other signals for the family being Roma (Váradi, 2012). The selection of the names was based on previous research (Váradi 2012; Simonovits and Kézdi 2016); using combinations of Roma sounding first and surnames as we have already done in our previous experimental studies (Simonovits et al. 2022; 2023). We used the same names for the mothers across both sport fields in the emails inquiring about the trial practice and included girl and boy names for their children according to the type of sport. In the emails the combinations of the two names were fixed; i.e. the same pairings of mothers’ and children’s names were used throughout the studies. We created separate Gmail accounts for all mother names, altogether 8 accounts. (See Table S2 in Supplementary materials on the email profiles and names).

Following Storm et al. 2023, we categorized responses that did not indicate an explicit rejection of the child to take part in the trial practice as positive responses. Following previous experimental studies (Gomez-Gonzalez et al. 2021; Nesseler et al. 2019. and Storm et al. 2023) we categorized both negative and non-responses as negative responses. We used this binary categorization of discriminatory behavior in the descriptive and explanatory models of our analysis. Beyond this simple binary categorization of the outcome variable, we intended to further analyze the content of the responses to explore potential subtle forms of unequal treatment. Responses were coded by two independent coders, according to multiple criteria. See the coding protocol in Appendix 1.

Results

In this section we present the descriptive results, followed by the estimation of receiving positive responses by ethnicity and gender (in the field of basketball) through logistic regression models.

The proportion of positive responses by ethnicity and sport is presented in Figs. 1 and 2 (for further details on the number of observations by subgroups see Table S3 in the Supplementary Materials). We found remarkably high positive response rates in both sports, and we did not find significant differences in the response rates sent to Roma and non-Roma mothers in either of the two sports (Table 1). The gap in the positive response rates between the Roma-sounding names compared to non-Roma sounding names was 2 percentage points in the case of volleyball (Fig. 1 panel A) and 4 percentage points in the case of basketball (Fig. 1 panel B). Gender difference was found to be statistically significant (see the Welch Two Sample t-test, in Table 1) i.e. overall mothers applying with daughters got positive answers at a higher rate (compared to mothers with sons) in the basketball sample.

Fig. 1: Percentage of positive replies by ethnicity and by type of sport in study 1, with Confidence Intervals (CI).
figure 1

A Volleyball (N = 120). B Basketball (N = 150).

Full size image
Fig. 2: Percentage of positive replies by gender and ethnicity in basketball in study 1, with Confidence Intervals (CI).
figure 2

A By gender – Basketball. B By gender and ethnicity – Basketball.

Full size image
Table 1 Response rates by ethnicity in study 1, with T-testsa.
Full size table

Variance in response rates by names

Certain Roma and non-Roma name combinations received higher levels of positive responses than others; however, name-specific differences in response rates were inconsistent across the two sports. Statistical tests (Chi-squared tests run separately for Roma and non-Roma profiles within both fields) did not show significant differences within the profiles; however, it might be the case of Type 2 error, as the number of observations by names were low (see Table S3 in Supplementary Materials for Study 1).

Logistic regression models

We applied binary logistic regression models to test whether Roma-sounding names were less likely to receive positive replies than their non-Roma counterparts (Tables 2 and 3).

Table 2 Results of the logistic regression models, Beta, (S.E) [Exp Beta]a, in Study 1.
Full size table
Table 3 Response rates by ethnicity and sport, with Welch two sample T-testsa.
Full size table

In line with the descriptive results, we found non-significant differences between the treatment of the families with Roma and non-Roma sounding names in the fields of volleyball and basketball. Gender, which was used as a secondary experimental variable in the basketball sample, had a significant and negative effect on response rates (at the 5%-level): mothers with sons had lower odds of receiving a positive response compared to mothers with daughters, controlling for ethnicity. When the interaction term was included in the logistic regression model (see row 4 in Table 2), its effect was not statistically significant i.e. the measured gender differences do not differ across the Roma and the non-Roma subgroup.

Further analysis of responses

To explore more subtle forms of unequal treatment, we analyzed responses according to tone (politeness and formality) and content (informativity and giving further tasks to the mother). In line with the earlier results based on binary outcome (positive response rates by ethnicity and gender), we did not find significant differences when applying a more nuanced analysis of the responses (See Appendix 1 on the coding protocol of the answers). In other words, none of the analyzed aspects of the answers’ tone (politeness and formality) or any of the content dimensions (informativity and giving further tasks to the mother) differed significantly by ethnicity.

Discussion

In study 1, we observed high positive response rates in both volleyball and basketball, yet no significant disparity in responses comparing mother with Roma- and non-Roma sounding names in either of the two sports (Fig. 1). In the basketball sample, where we included both daughters and sons in the inquiries, mothers with daughters received higher rate of positive responses compared to mothers with sons. This gender difference was found to be statistically significant, regardless of ethnicity. The reasoning behind this may be found in the infrastructure of the youth categories and teams within the Hungarian Basketball Federation. For kids aged 11, there are 13 regions in total (targeting different parts of Hungary), each including 8 teams (MKOSZ 2023). These regions are divided into boys, girls, and mixed teams. However, out of all the regions, only 4 have teams exclusively for females. The rest of the regions have teams exclusively for males or mixed teams. The mixed teams usually have 1 or 2 female players (MKOSZ 2023). This gender gap is also evident in other age categories, including older youth groups and professional basketball teams. Hence, female players are always welcome to join with the aim to further develop young female talents and the women’s basketball in general in Hungary.

Despite the general trends of discrimination towards Roma people in Hungary (Csomor et al. 2021; Simonovits et al. 2022; Mikula and Montag 2023) across various domains, in this study we found no significant differences. Even though we tested the full population of volleyball and basketball clubs, according to our power analysis, the numbers of observation might not have been sufficient to detect discrimination (similar to the rates found in of previous studies). As we only tested the prevalence of discrimination, we do not have sufficient information regarding mechanisms that hindered or mitigated discrimination of Roma children applying for basketball and volleyball practice. One possible explanation for not having found discrimination is the fact that Roma mothers inquiring through email may have been perceived as non-stereotypical, indicating higher social class through the language of the inquiries and the fact that they had access to internet. As discussed earlier, there is a prevalent stereotype about Roma people being poor and having many children to gain benefits from the state.

Study 2

In Study 2, we tested discrimination of Roma children of low social class seeking participation in an under-8 soccer trial session.

Methods and materials

Study 2 followed an identical design and procedure as Study 1, with the following modifications. Given that there was a sufficiently high number of soccer clubs in the U8 age group, in contrast to study 1, this study had a between-subject (unpaired) design. This way we could minimize the risk of detection and minimize the burden on our subjects. In order to examine the interplay between ethnicity and social class, soccer clubs were contacted using the same email accounts of the mothers as in Study 1, having Roma-sounding names coupled with a strong signifier of low socioeconomic status. Mirroring the male-domination of soccer, only emails by mothers with sons were used in the experiment.

Participants

We identified and included all soccer clubs offering training opportunities for children aged under 8 years, i.e. the youngest age group for soccer training. These soccer clubs were partly selected out of the database provided by the research group of Gomez-Gonzalez et al. (2021), partly from a Hungarian websiteFootnote 6 collecting all soccer clubs with an online presence.

Procedure

The procedure of Study 2 exactly followed that of Study 1, using the same names and a similar email content, with the addition of the cue about stereotypical poverty of Roma by indicating that the mother has five children. Emails were sent out in March 2023, i.e. outside of regular registration period.Footnote 7Footnote 8

Version 1 – Roma profile:

Subject: Trial training

Dear Coach!

I don’t know if I’m writing to the correct address, but my family and I are moving to your area soon and I would like to ask whether it is possible to attend a training session at your club. If you are not the person responsible for this, please forward this email to the coach. Thank you very much!

I’m [mother’s full name], mother to five children, and soccer is a great love of our family. My eldest son, [son’s first name], is now 7 years old (born 2015) and we would like for him to take football more seriously.

Best regards:

[mother’s full name]

Version 2 – non-Roma profile:

Subject: Trial training

Dear Coach!

I don’t know if I’m writing to the correct address, but my family and I are moving to your area soon and I would like to ask whether it is possible to attend a training session at your place. If you are not the person responsible for this, please forward this email to the coach. Thank you very much!

I’m [mother’s full name] and football has always been a great love of our family. My eldest son, [son’s first name], is now seven years old (born 2015) and we would like football to play a bigger role in his life.

Best regards:

[mother’s full name]

In Hungary, Roma families on average have a higher number of children compared to non-Roma families, though there is no reliable information regarding the exact number of children in Roma families due to the difficulties and controversies of collecting ethnic data (Messing and Pap 2024). In our study, we used the information about having high number of children to activate a common stereotype about Roma families having many children for the sake of receiving family benefits as a substitute for income (Enyedi et al. 2005; Halász and Fejes 2024). Responses and non-responses to the emails were recorded within the period of 11 days. In case of any response, a polite email informing the club that the child is no longer interested in participating in the practice for practical reasons was sent within 9 days. All participants of the research were debriefed within two months (see Appendix 3).

Measures

Measures were identical to those in Study 1 and the same codebook was applied to the coding of the content of the responses to the email inquiries.

Results

Descriptive statistics

In this section we present the descriptive results of Study 2, followed by the estimation of receiving positive responses by ethnicity through logistic regression models.

Answering the main question of Study 2, when Roma ethnicity was coupled with a stereotypical signifier of low social class, we found a significant ethnicity gap based on the comparison of positive responses received by Roma low status and non-Roma mothers (Table 4). We found that requests sent out by mothers with non-Roma sounding names received almost twice as many answers than the ones sent out by Roma sounding names signifying low status (response rates of 72% vs. 39%) (Fig. 3).

Table 4 Response rates by profiles (non-Roma vs Roma mother names) in study 2, Pearson’s Chi-squared test.
Full size table
Fig. 3
figure 3

Percentage of positive replies by ethnicity, in Study 2 (N = 190), with Confidence Intervals (CI).

Full size image

Variance in response rates by names

Comparing the positive responses by names, we also found large and statistically significant within-group differences in the case of the non-Roma profiles; i.e. one of the mother names (Molnár Lilla) received an extremely high proportion (91%) of positive responses, compared to the others, however, we cannot interpret this difference, as there is a lack of empirical study on name perception. In contrast, within group differences were not significant across the Roma names. However, the children’s names may also have influenced the coaches’ perception of ethnicity, making it uncertain what exactly caused these within-group differences.

Logistic regression model

In accordance with the descriptive statistics, writing an email with a Roma sounding name and a stereotypical signifier of low social class, have a negative and significant impact (at the 1% level) on receiving a positive response in our model: our model estimated an odds ratio of 0.25 for non-Roma vs Roma boys to be invited for the trial practice (i.e. the odds receiving a positive response as the mother of a non-Roma boy is 4 times more likely to receive a positive response than the mother of a Roma boy) (Table 5).

Table 5 Results of the logistic regression model, Beta, (S.E) [Exp Beta]a in Study 2.
Full size table

Further analysis of responses

In line with the analysis above on the main outcome variable (positive response rates by ethnicity), we found significant differences in all of the dimensions i.e. both of the analyzed aspects of the answers’ tone (politeness and formality) and the content dimensions (informativity and giving further tasks to the mother) was significantly different (Figs. 4 and 5; Tables S4–S5 in Supplementary Materials).

Fig. 4: Response tone in percentage by ethnicity – Soccer (no responses excluded).
figure 4

Pearson’s Chi-squared test, ***p < 0.001, **p < 0.05; *p < 0.10. Chi-squared = 16.849, df = 2, p-value = 0.0002195 ***.

Full size image
Fig. 5: Additional tasks for the mother in percentage by ethnicity – Soccer (no responses excluded).
figure 5

Pearson’s Chi-squared test Chi-squared = 13.732, df = 3, p-value = 0.003294 ***.

Full size image

Analysing the results excluding non-responses, we observed significant differences by ethnicity in one of the tone-related and in one of the content-related aspects. Based on the Adjusted Residuals (in Tables S4–S5 in the Supplementary Materials) we found that Roma mothers received ‘colder’ responses (i.e. with a dry tone) more frequently than their non-Roma counterparts, while non-Roma mothers were more likely to receive neutral responses at a higher rate, however in terms of the cordial response (i.e. with encouraging words) type we did not find significant difference by ethnicity (Fig. 4). When comparing the task related instructions given by the coaches, Roma applicants received significantly fewer of these types of inquiries compared to non-Roma senders, which may be interpreted as coaches tend to request additional relevant information from non-Roma applicants at a higher level than from Roma mothers (Fig. 5).

Discussion

In Study 2, requests from non-Roma sounding names received nearly twice as many responses as those from Roma sounding names signifying the stereotypically low social class of the mother (72% vs. 39%). Furthermore, significant differences were detected across other, more nuanced, outcome variables indicating subtle forms of discrimination, including both content and tone related indicators.

General discussion

Results of the two field experiments show that when Roma ethnicity was only signified by names (Study 1, basketball and volleyball), no different treatment of Roma applicants occurred. Meanwhile, when in addition to Roma-sounding names a stereotypical signifier of lower social class was present, discrimination was found to be strong (Study 2, soccer). This is a possible indication of the intersectionality of ethnicity and social class (Urbiola et al. 2022). Furthermore, the unequal treatment of Roma applicants with low social status was found to be present both in its blatant and subtle forms (Jones et al. 2016). First, Roma applicants with a low social status had a one to four chance of getting a positive response compared to non-Roma applicants with higher social status. At the same time, some of the content and tone related indicators of the responses in Study 2 also differed, indicating more welcoming attitudes towards non-Roma / upper-class applicants. These findings mirror the earlier findings of an experimental study assessing discrimination against Roma people by Hungarian local governments (Buda et al. 2023).

Thus, our results imply a possible intersection of ethnicity with poverty triggering discrimination (Cho et al. 2013; Settles and Buchanan. 2014). This might be based on two parallel processes. The negative treatment of lower-class applicants is coupled with the activation of negative stereotypes about Roma people being lower class (Urbiola et al. 2022; Hester et al. 2020). In Study 1, the fact that Roma applicants were writing emails in a language signifying higher educational level and higher socio-cultural status, may have disconfirmed stereotypes and thus, led to the lack of unequal treatment (Hewstone 1989). This effect was countered by activating a strong stereotype in Study 2. Results are in line with the literature on stereotype activation, suggesting that in case stereotypes about an outgroup member are activated, chances of expressing prejudice and negative treatment are higher (Kunda and Spencer 2003). Also, the stereotypical information itself may have drawn attention to the name, used as the ethnic trope, that otherwise might have been overlooked (Abramova 2020).

It is important to point out that since we were not able to include control groups of non-Roma low-status applicants and Roma high-status applicants due to the limited size of the population we tested, an important question remains unanswered. Our data does not provide a clear answer to the question whether the discrimination we uncover in the field of soccer is based on ethnicity, lower social class, or both. Therefore, Study 2 should be seen as a pilot, paving the way for further studies aiming to disentangle these two effects.

Our findings of Study 2 are also consistent with findings of previous research on male applicants to soccer clubs, consistently showing that ethnic minorities experienced discrimination (Gomez-Gonzalez et al. 2021; Nesseler et al. 2019; Nesseler et al. 2023). We found that discrimination against lower-class Roma applicants is even more pronounced (30%) than discrimination against migrant populations in Hungary without indicating their social class (20%) (Gomez-Gonzalez et al. 2021).

Compared to prior studies employing a similar methodology (Storm et al. 2023; Nesseler et al. 2023), our findings corroborate the observation that female applicants, irrespective of ethnicity, receive notably higher response rates than their male counterparts. The pattern of not finding significant differences in responses based on ethnicity among female applicants aligns with Storm and colleagues’ (2023) research in Denmark and Norway; however, it contrasts with Storm’s (2023) findings in Sweden and Nesseler and colleagues' (2023) investigation in Italy. Notably, our study focused on female applicants in basketball and volleyball, unlike prior studies which primarily examined soccer.

Conclusions, limitations, and future research

As we applied online correspondence tests in our studies, we were able to capture discrimination only at the very first step of the process of Roma children’s participation in sport. Therefore, our findings could be understood as the most conservative estimation of discrimination, as there are several further hurdles minority children might face even if the initial response received by the mother is not negative. Having followed the design of previous studies testing ethnic discrimination in sport, we used emails as the channel of communication. It should be noted, however, that this may not be a widely accessible option for all Roma mothers, due to financial and educational barriers.

Additionally, all correspondence studies suffer from post-treatment bias (Coppock 2019; Buda et al. 2023), due to varying levels of nonresponse rates found by ethnicity. As our study revealed a higher non-response rate towards Roma mothers, comparing response quality indicators based on the responses may lead to post-treatment bias. It is plausible that many coaches with anti-Roma attitudes choose not to respond to Roma mothers’ inquiries, therefore an under-estimation rather than an over-estimation of discrimination more likely emerged from the response analysis in our studies. This idea is supported by the significantly higher levels of significance observed when including non-responses in the categorical analysis, compared to when focusing solely on the responses received (see the different Chi square statistics in the Supplementary materials for Study 2). Finally, it has to be noted, especially in Study 1, that the number of observations by subcategories were relatively low, compared to previous experimental studies in the field.

Moreover, it is important to acknowledge that besides the intersectionality of social class and ethnicity, the different observed sport fields may have also influenced our findings, potentially introducing bias. Discriminatory practices towards the Roma community at the entry level were notably less prevalent in basketball and volleyball compared to soccer. However, the specific factors driving this variance, including for instance the dependence to recruit underaged players to access TAO funds and disparities in coaches’ educational backgrounds, remain unidentified within the scope of this study.

To the best of our knowledge, our study is the first attempt to assess discrimination in the field of sport against Roma people. In contrast to prior experimental studies in this domain, our research broadens its scope by encompassing additional types of sports beyond soccer, and by adopting an intersectional approach to measuring discrimination.

Regarding the generalizability of our findings, we acknowledge the limitations imposed by our study’s sample size, geographical scope, and focus on three selected sports fields. In alignment with Storm et al. 2023, we advocate for the exploration of various other sports, beyond the most popular team sports, as potential arenas for social integration of ethnic and other minorities. We believe our study sheds light on discriminatory practices against Roma children in sports, thereby paving the way for future research in this direction.