Ethnic and gender discrimination in recruitment: experimental evidence from Finland

We ask (1) how the position of an ethnic (majority or minority) group in the local ethnic hierarchy affects the amount of recruitment discrimination faced by applicants from that group, and (2) whether gender discrimination is dependent on occupational gender stereotypes in the same way among ethnic majority and minority applicants. We use the situation testing method for the first time in Finland: In an experimental study (Study 1), 103 dentistry students made recruitment decisions based on the CVs of three bogus applicants from different ethnic groups (Finnish, Austrian and Polish) and in a field experiment (Study 2), four test applicants (male and female Finns and Russians) with equivalent CVs applied for 1,258 vacant jobs, addressing gender discrimination in relation to occupational gender stereotypes as well as ethnic discrimination. Together these studies cover both skilled (Study 1) and semi-skilled jobs (Study 2) and applicants from ethnic minority groups originating from within as well as outside the EU. Results show that majority group members are more likely to be hired compared to minority members (both Studies) and that minority members from a higher status group are more likely to be hired than those from a lower status group (Study 1). Results also show that male applicants from the majority group were discriminated compared to women in occupations characterised as feminine, while Russian men faced recruitment discrimination compared to Russian women independently of the job’s gender stereotype (Study 2). Implications of recruitment discrimination based on ethnicity and gender are discussed.

ation. However, the debate continues on the extent to which social and economic inequality between different groups in society is due to discrimination. Differences observed in the labour market outcomes (wages, employment rate, etc.) may be the result of discrimination but also of differences in productivity. Economists traditionally define labour market discrimination as a situation in which equally productive persons are treated unequally on the basis of an observable characteristic (Laing, 2011). From a social psychological perspective, recruitment discrimination can be defined as a reduced likelihood of being offered a job or job interview following submission of an application to an advertised vacancy when the applicant's membership in a negatively stereotyped group is revealed in the application (Stone & Wright, 2013). Our first research question addresses the extent to which the position of an ethnic group in the local ethnic hierarchy-as a majority group or a high or low status minority group-affects the amount of recruitment discrimination faced by applicants from that group, even if their qualifications match those of applicants from ethnic groups higher in the hierarchy.
While there is some evidence showing that membership in a devalued ethnic group can indeed be the reason for being discriminated against during the recruitment process (Krings & Olivares, 2007), the role of the applicant's gender still remains unclear. Even less is known about whether or not ethnic discrimination follows similar patterns for male and female members of a particular ethnic group. So far, research has shown that immigrant men may sometimes face more recruitment discrimination than immigrant women (e.g., Bertrand & Mullainathan, 2004).
However, immigrant women may also be more discriminated than immigrant men, as it is assumed that recruitment discrimination on the ground of gender may result in intersectional discrimination: Minority women may suffer from a double stigma due to both ethnicity and gender (e.g., Petit, Duguet, L'Horty, du Parquet, & Sari, 2011).
One reason for these contradictory results on gender discrimination may be that also occupational gender stereotypes, that is, whether a job is considered stereotypically feminine or masculine, have been found to predict discrimination faced by men and women in the recruitment situation (e.g., Carlsson & Rooth, 2008). The likelihood of discrimination in recruitment is affected not only by gender and the position of an ethnic group in the local ethnic hierarchy (i.e., the degree of social acceptance of that group among the population) but also by whether or not the stereotypes related to the applicant's gender and/or ethnic group are congruent with the occupational stereotype (Weichselbaumer, 2004).
Discrimination based on ethnicity and gender, while acknowledging the gender stereotype of the occupation, is still under-researched, and results are inconclusive. One example is the study by Arai, Bursell, and Nekby (2011).
The researchers tested gender discrimination of Arab-named and Swedish-named male and female job applicants in Sweden but found mixed evidence on the effect of occupational gender stereotypes. Specifically, Arabic women were invited to an interview more often than Arabic men for the stereotypically "feminine" accountant and assistant nurse positions but also for the post of a "masculine" computer specialist. Arabic men, in turn, had better chances of being hired for a "masculine" driver position but also for "feminine" high school teacher positions. Our second research question therefore concerns the extent to which gender discrimination in recruitment is dependent on occupational gender stereotypes in the same way among ethnic majority and minority applicants.
To answer these two research questions, discrimination should be measured accurately. This is, however, a challenging task (e.g., Bond, McGinnity, & Russell, 2010). The major benefit of (laboratory and field) experimental research is the possibility to control and manipulate the research design, which makes it possible to test the treatment of two equally merited candidates of, for example, different ethnic background (Aalto, Larja, & Liebkind, 2010;Pager, 2007;Quillian, 2006). In Finland, however, a country characterised by a short history of immigration and a concomitant small number of immigrants, no systematic research on recruitment discrimination of immigrants has been carried out before.
In this paper results will be reported from one laboratory experiment (Study 1) and one field experiment (Study 2) in Finland. Study 1 addresses the first research question as it focuses on recruitment discrimination on the basis of ethnicity and the position of ethnic (majority and minority) groups in the local ethnic hierarchy. Discrimination in recruitment towards members of low status (Polish) and high status (Austrian) ethnic minority groups from within the European Union (EU) will be investigated, comparing them to each other as well as to the Finnish majority group. In a between-subjects design, three CVs of bogus applicants representing the same highly skilled profession (dentists) with similar skills and work experience but with names and other information revealing their ethnic group membership are presented in pairs of two to dentistry students who are asked to make a recruitment decision. Study 2 focuses on the role of both ethnicity and gender in recruitment discrimination occurring in occupations characterised by different occupational gender stereotypes and thus addresses both research questions.
In Study 2, discrimination in recruitment, compared to Finnish majority applicants, towards men and women from a low status ethnic minority group (Russians) from outside the EU will be investigated by sending fake applications to vacancies in male-and female-dominated semi-skilled jobs, controlling for skills and work experience in the applicants' CVs.
To our knowledge, field experiments of recruitment discrimination of Russian-named job applicants have not been conducted in any EU country before. In addition, as noted above and as will be further evident below, previous research on discrimination based on ethnicity and gender, taking into account the gender stereotype of the occupation is scarce and the obtained results are inconclusive. Together our studies add to the existing knowledge on the recruitment discrimination of rarely studied ethnic groups in a new social context by covering both skilled (Study 1) and semi-skilled jobs (Study 2), thus representing a more diverse domain for discrimination in recruitment compared to previous studies (e.g., Arai et al., 2011). A further contribution is that we explore how the compatibility of the applicants' gender with the occupational gender stereotype affects the amount of recruitment discrimination faced by ethnic minority and majority job applicants.

Recruitment Discrimination: Ethnic Hierarchies and the Recognition of Professional Qualifications
So called situation tests (for a review, see e.g., Riach & Rich, 2002), where a pair of equally qualified job applications with different names are sent in response to real job advertisements, have shown systematically high levels of recruitment discrimination of various groups in society on the grounds of, for example, ethnicity, nationality or gender in Europe (e.g., Carlsson & Rooth, 2007;McGinnity & Lunn, 2011), the United States (e.g., Bertrand & Mullainathan, 2004) and Australia (e.g., Booth, Leigh, & Varganova, 2010). Currently a widely accepted view of any prejudiced behaviour, like recruitment discrimination, is that it can operate automatically (without conscious intent) or systematically (with conscious intent), and that this corresponds to explicit (deliberate) and implicit (automatic) attitudes, respectively (e.g., Agerström & Rooth, 2009. Although recruitment discrimination has been shown to take place also without conscious intent (e.g., Agerström & Rooth, 2009Rooth, 2010), regardless of its underlying mechanisms it represents unjustifiable negative behaviour towards a group or its members on the basis of their group membership (Al Ramiah, Hewstone, Dovidio, & Penner, 2010). One's membership in an ethnic, national or gender group, in turn, can be easily detectable from the individual's name Although membership in an ethnic and national group has been found to predict recruitment discrimination (e.g., Bertrand & Mullainathan, 2004;Krings & Olivares, 2007), in this paper it is argued that it is not ethnicity / nationality as such which should be taken into account when labour market discrimination is studied but so called ethnic hierarchies. Ethnic hierarchy has been defined as a ranking of different minority groups in the country, in which a minority group's position is determined by factors such as perceived cultural characteristics of the group, its socioeconomic status and the perceived threat that it poses to the national in-group (Pepels & Hagendoorn, 2000).
Although ethnic hierarchies differ between countries, previous research has shown that in Western multicultural societies ethnic rankings seem to follow a certain pattern. Specifically, West Europeans are at the top of the hierarchy, South and East Europeans are in the middle and non-European groups occupy the lowest positions of these ethnic rankings (see e.g., Berry, 2006;Hagendoorn, 1993;Snellman & Ekehammar, 2005).
So far, the effect of position of a certain minority in an ethnic hierarchy on the amount of discrimination in recruitment has been largely undiscovered. There is some evidence showing that the position of an ethnic group in the hierarchy has consequences for its members' labour outcomes. Specifically, studies in the Canadian context (e.g., Thompson, 2000) corroborate that labour market outcomes of immigrants' are not predicted by their level of education. As showed by Bauder (2003), this finding is especially true for highly-skilled immigrants and it is often a consequence of a long and arbitrary procedure of the recognition of immigrants' professional qualifications. The procedure of the recognition of professional qualifications in Canada differs, however, depending on the origin of the qualifications.
While immigrants from Asia, Middle East and Eastern Europe have restricted access to high-skill occupations (Thompson, 2000) due to the arbitrariness of the recognition process (Bauder, 2003), immigrants from Western European and English-speaking countries have an advantage in the labour market (e.g., Gozalie, 2002) as their qualifications are often automatically recognised (see Girard & Bauder, 2007).
As showed by Berry (2006), Canadians view West and North European immigrants more positively than immigrants of East and South European origin, while immigrants of non-European background occupy an even lower position in the ethnic hierarchy. Therefore, the degree of difficulty in the recognition of professional qualifications seems to match the position a diploma holder occupies in the ethnic hierarchy in Canada: Immigrants whose ethnic group occupies a high position in the hierarchy do not struggle with the recognition of their qualifications, whereas immigrants from groups occupying low positions in the hierarchy are less often recognised as professionals. It has thus been claimed (Bauder, 2003;Girard & Bauder, 2007) that the non-recognition of qualifications is the reason for immigrants belonging to low status ethnic groups to be underemployed when compared to immigrants belonging to high status ethnic groups. As a consequence, in research on recruitment discrimination in receiving societies, the effect of the recognition of professional qualifications seems to be confounded with that of the local ethnic hierarchy. Thus, the effect of the recognition of qualifications has to be controlled for or eliminated when studying positions which require that an applicant' s professional status is recognised in the receiving state. Only in this way is it possible to corroborate the effect of ethnic hierarchies on labour market discrimination.
In the context of the EU, literature (e.g., Currie, 2008) indicates that East and Central European immigrants (A8 i group nationals) have a lower status in the labour market than immigrants from the Western and South European countries (EU15). According to Anderson, Ruhs, Rogaly, and Spencer (2006) this is indicated by, for example, a clear mismatch between skills and qualifications reported by East and Central European immigrants and skills needed for a successful performance of the (mostly unskilled) jobs they hold in the UK. In addition, these immigrants benefit significantly less from their human capital than their counterparts from the Western and South European countries (EU15), even after controlling for personal characteristics (Anderson et al., 2006). Also in Finland, im- migrants from East and Central Europe (A8) are sometimes positioned in the official discourse as a remedy for the lack of local workers in the least attractive and prestigious sectors of the Finnish economy as, for example, elderly care (see Jaakkola, 2005).
Most studies on recruitment discrimination against members of ethnic minorities have taken place in countries with fairly large numbers of immigrants and a long history of immigration. One exception is the study by McGinnity and Lunn (2011), who found no differences in recruitment discrimination against Africans, Asians and Germans in Ireland. The authors explain this result as being due to the fact that the recent immigration history of Ireland has prevented stereotypes specific to various immigrant groups to develop. Contrary to the situation in Ireland, however, in Finland a consensual hierarchy of immigrant groups does exist (Jaakkola, 2005(Jaakkola, , 2009, cf. Context of the Present Studies below). This study, therefore, contributes to existing knowledge on recruitment discrimination by investigating the largely undiscovered effect of the ethnic hierarchy of a small country with a short history of immigration on the amount of discrimination in recruitment in that country. Based on the reviewed literature it is expected that, in the recruitment process, majority group members are preferred over ethnic minority groups of different status in this hierarchy (Hypothesis 1A) and that minority groups of higher status are preferred over those in a lower status, regardless of their qualifications or the formal recognition of those qualifications (Hypothesis 1B). H1A will be tested in both Study 1 and Study 2 while H1B will be tested in Study 1.

Occupational Gender Stereotypes and Intersectionality
Significant, persistent and pervasive levels of discrimination against women in the labour market have been repeatedly reported in reviews on relevant research (e.g., Riach & Rich, 2002). However, studies on gender discrimination have shown the importance of the gender stereotype of the occupation on the amount of discrimination: Being a woman is either an advantage or disadvantage, depending on whether the occupational stereotype is more masculine or feminine (Carlsson & Rooth, 2008;Nunes & Seligman, 2000). In Study 2, we thus assume that among majority members, female applicants will be discriminated when applying for stereotypically masculine jobs and male applicants when applying for stereotypically feminine jobs (Hypothesis 2).
As there are no ethnic minority (or majority) members without gender, gender discrimination can be expected to occur in the same way among members of ethnic minorities and majorities. However, results concerning the effect of occupational gender stereotypes are not directly generalisable to men and women belonging to ethnic minorities.
Studies (for a review, see Arai, Bursell, & Nekby, 2008) show that stereotypes about different ethnic groups often reflect stereotypes about men belonging to that group, whereas ethnic minority women may be stereotyped very differently from what the stereotypes of their ethnic group would suggest, and they may be stereotyped simply according to the general female stereotype.
In line with these findings, Ghavami and Peplau (2013) corroborated that cultural stereotypes of an ethnic group were indeed more similar to stereotypes of the men than the women from that group. Nevertheless, the authors also noticed that culturally held intersecting gender and ethnic stereotypes contained unique elements that could not be obtained by simply adding gender stereotypes to ethnic stereotypes, and that the stereotype of minority women contained more distinct elements than that of minority men. There is some evidence that women from minority groups suffer from a double stigma caused by their gender and ethnicity. For example, Petit et al. (2011) found that Moroccan and Senegalese women were discriminated against more than Moroccan and Senegalese men and more than native French men and women. However, Bertrand and Mullainathan (2004) found the opposite, that is, African American women encountered less discrimination than African American men, and the same results were reported also by Petit et al. (2011) for Vietnamese women and men. In addition, Martens et al. (2005) found no difference in the amount of discrimination faced by minority women and men and Arai et al. (2011) initially found more ethnic discrimination against Arabic women than Arabic men, but after the minority CV's had been made better than those of the Swedish native applicants, only Arabic men were discriminated. Therefore, the evidence on double stigma as opposed to single effects of ethnicity or gender is very mixed.
One possible explanation for these mixed results could be the occupational gender stereotypes of the studied occupations. Besides ethnic hierarchy, the likelihood of discrimination is affected by the overall stereotype match, that is, whether or not the (ethnic and/or gender) stereotype of the applicant is congruent with the occupational stereotype (Weichselbaumer, 2004). For example, women are often considered more suitable for service jobs and men for driving or construction. However, studies showing how gender and ethnicity affect the discrimination faced by ethnic minority men and women when applying for stereotypically 'feminine' and 'masculine' jobs are scarce. For example, when Petit et al. (2011) studied the interaction of ethnicity and gender on discrimination in a highly-skilled job for IT-developer, they held the occupation (IT-developer) constant. As a consequence, it cannot be judged whether or not their results are specific to this occupation only, as the stereotype match is occupationspecific (Weichselbaumer, 2004).
The few studies conducted to date on discrimination based on both ethnicity and gender which also acknowledged the gender stereotype of the occupation have produced inconclusive results. Studying Moroccan job seekers in Brussels, Smeesters and Nayer (1998) found that Moroccan men were discriminated against compared to women in the hotel and restaurant sector and in non-sales services; jobs in both of these sectors may be considered stereotypically feminine occupations. In the more gender-neutral sales and business jobs, the authors found hardly any evidence of gender discrimination among Moroccan job seekers. Although Arai et al. (2011) found some support for the occupational gender stereotype hypothesis, their results were mixed. Specifically, Arabic women were invited to an interview more often than Arabic men for some feminine, but also for some masculine positions, while the same was true for Arabic men regarding other masculine as well as feminine positions. Thus, there is clearly a need for more research in this area. Due to the inconclusive results obtained in previous research we will, in Study 2, only explore whether or not the treatment of male and female ethnic minority (Russian) applicants in stereotypically masculine and feminine occupations follows the same pattern as that received by ethnic majority (Finnish) male and female applicants.

Context of the Present Studies
Like some other small countries in Northern Europe, Finland has traditionally been rather a country of emigration than immigration (Pitkänen & Kouki, 2002). Still today, only five per cent of the population speaks a language other than Finnish (90% of the population) or Swedish (5%) as their mother tongue. Due to historical reasons ii , the population of Finland is quite patriotic and shows rather high levels of national identification and pride in survey polls (Finell, 2012). The content of Finnish national identity is also quite essentialist, i.e., builds upon ethnic and cultural characteristics rather than citizenship (Varjonen, Arnold, & Jasinskaja-Lahti, 2013). Thus, the Finnish context is characterised by a relative lack of experience of ethnic or cultural minorities, combined with high levels of a cohesive national identity.
The consensual hierarchy of immigrant groups in Finland has remained quite stable over the years (Jaakkola, 2005(Jaakkola, , 2009 , and 24 th position, respectively) occupy the lowest positions in the hierarchy (Jaakkola, 2005, p. 138). The reasons for the negative stereotypes of Russians may be sought in a conflictual history between the former Soviet Union and Finland.
In Finland, no systematic research on employment discrimination among ethnic minority groups has been conducted, but unemployment is more frequent among some immigrant groups than among Finnish majority members. For example, along with refugees, immigrants from Russia and the former Soviet Union have been worst affected by unemployment (25% of Russian speakers in the labour force were unemployed in 2011), as compared to, for example, only 10% of unemployed among Estonian citizens (Statistics Finland, 2013a). Research results also indicate that Russians may encounter intentional or unintentional discrimination in the labour market (e.g., Jasin-

Study 1
The aim of Study 1, conducted as a laboratory experiment among Finnish university students, was to answer the first research question by examining whether Finnish majority applicants are preferred over minority applicants belonging to ethnic groups of different status in the local ethnic hierarchy when the qualifications of these applicants are the same (Hypothesis 1A). In addition, it was investigated whether applicants belonging to minority groups of higher status are preferred over those of a lower status (Hypothesis 1B). Specifically, we wanted to find out whether the status of the specific EU nationality (high status immigrant applicant from Western and South Europe vs. low status immigrant applicant from Central and Eastern Europe) contributes to recruiters' decisions when selecting a successful candidate for a professional position between (a) a Finnish majority applicant and an ethnic minority applicant, and between (b) two ethnic minority applicants from within the EU holding different positions in the local ethnic hierarchy but not differing from each other with regard to the recognition of their professional qualifications.
When selecting the high status job requiring professional skills to be studied, the possible interference of the nonrecognition of professional qualifications with the recruitment outcome was eliminated by the selection of the dentist profession. As dentistry training within all EU member states is unified with respect to its content, dentists trained inside the EU have the right to claim automatic recognition of their qualifications under the specific sector rule (see Directive 2005/36/EC of the European Parliament).
When selecting the ethnic minority groups for this study two kinds of nationalities were avoided: those which are either especially privileged in Finland (e.g., English-speaking, Nordic, Estonians) or those which could potentially elicit strong negative attitudes due to salient economic or political crisis at the time of the study (e.g., Greece, Spain or Portugal). Instead, countries where the length of the dentistry education and post-graduate internships were structured similarly to those in Finland, and from where immigration to Finland would be of sufficient magnitude to be realistic, were preferred. The two countries which fit best all of these criteria were Austria and Poland, the former representing a high status EU country (Western and South Europe) and the latter a low status EU country (Central and Eastern Europe).
We expected that a low status ethnic minority applicant (from Poland) will be discriminated against when paired with a high status majority Finnish applicant and that even a high status ethnic minority applicant (from Austria) will be discriminated when paired with a majority Finnish applicant with equivalent qualifications (Hypothesis 1A).
We also expected that an ethnic minority applicant (from Austria) with a higher position in the ethnic hierarchy will be favoured over a low status ethnic minority applicant (from Poland) with equivalent qualifications (Hypothesis 1B).

Method
The data was collected in 2011 among Finnish dentistry students at the Institute of Dentistry of the University of Helsinki. The final sample was N = 103 (M age = 24.57, 71.8% female). The percentage of women in the sample roughly corresponds to the percentage of females in the general population of dentistry undergraduate program students in Finland (74%; Kravitz & Treasure, 2009, p. 31).
The material consisted of three CVs of bogus male dentist applicants. The main difference between the CVs was the applicants' ethnicity (national origin) and the country of origin of their professional qualifications. Both the Austrian and the Polish CV contained information on the applicant's professional qualifications having been recognised by the relevant Finnish professional body (VALVIRA), and on the applicant being fluent in the national languages (Finnish and Swedish). Participants were informed that the purpose of the task was to investigate the willingness of Finns to employ candidates from different EU countries and that in their individualised study pack they would encounter CVs of applicants of different nationalities. As the presence of different EU nationalities in the task was made explicit, no manipulation check was administered to assess whether the participants noticed different ethnicities of the bogus applicants. The ethnicity of the applicants was also signalised by their name, mother tongue, and the location of their home university and previous work places in their home country.
In the task description the participants were asked to make a recruitment decision and select one of two available applicants for a vacant full-time position in a private dental clinic. The clinic was described as well established in the local dental care market, doing fine financially and presently employing ten additional dentists. In the betweensubjects design the CVs of two applicants with different nationalities were randomly handed out to the participants, creating three research conditions:

Results and Discussion
All hypotheses of the study were examined with chi-square tests. In line with Hypothesis 1A, the participants were significantly more willing to employ the high status majority Finnish than the low status Polish minority applicant (1) = 0.257, p = .612). Therefore, Hypothesis 1A received only partial support.
The results also showed that the participants were significantly more willing to employ the high status ethnic minority applicant from Austria than the low status ethnic minority applicant from Poland (Austrian: N = 28, Polish: (1) = 14.235, p < .001). Thus, Hypothesis 1B, which tested the effect of the position of the ethnic minority applicants in the ethnic hierarchy while controlling for recognition of professional qualifications, was supported iii .
The results of Study 1 indicated that the position of a particular ethnic minority group in the local ethnic hierarchy, even within the EU, may be a decisive factor between employment and unemployment, regardless of the official recognition of the group members' professional qualifications. However, Study 1 addressed discrimination within only one high-skilled profession against members of ethnic minority groups from two EU countries who do not form any substantial proportion of the immigrant population in Finland, and it did not address gender as a possible ground for recruitment discrimination. In addition, as laboratory experiments cannot demonstrate the true prevalence of recruitment discrimination due to not being based on random samples or total data, the results from Study 1 cannot be generalised (Aalto et al., 2010).

Study 2
Study 2 aims to validate the findings on ethnic discrimination (Hypothesis 1A) from Study 1 in a real-world setting and with an ethnic minority from outside the EU. Another aim of Study 2 was to answer the second research question by testing not only the effect of ethnicity, but also of gender as a ground for discrimination, which allowed for exploring also intersectional discrimination. To this end, Study 2 was set to assess whether among ethnic majority members female applicants will be discriminated when applying for stereotypically masculine jobs and male applicants when applying for stereotypically feminine jobs (Hypothesis 2), and to explore whether or not gender discrimination is dependent on occupational gender stereotypes in the same way for ethnic minority as for ethnic majority applicants.
Study 2 was a large-scale field experiment with real employers and vacancies. In this study, a research technique called situation testing was employed (Bovenkerk, 1992;ILO, 2007ILO, /2010. The basic principle of this testing method is that pairs of two equally merited fictitious applicants who differ only on the studied characteristic (e.g., ethnicity or gender) and otherwise are as similar as they can be without risking arousing the suspicion of employers, apply for the same jobs. If both applicants get the same treatment in the application process, e.g., both are invited for or both are denied an interview, no preferential treatment has occurred. However, discrimination is argued to occur if only one of them (usually the majority member) is repeatedly invited to an interview.
A field experiment on recruitment discrimination was conducted using the situation testing method which compared the success of fictitious male and female ethnic minority job seekers from outside the EU and equivalent job seekers from the Finnish majority group with matched education and work experience. Due to the limited scale of the Finnish labour market it was not possible to study simultaneously two ethnic groups and two grounds for discrimination. Thus, in order to have sufficient statistical power in our analysis, recruitment discrimination was studied on the basis of gender and one minority ethnicity when applying for semi-skilled office, restaurant, driver and construction jobs.
When choosing the ethnic group for Study 2, a number of conditions had to be met: 1) the studied group needs to be large enough for the results to be relevant; 2) members of the selected group should be typically employed in the semi-skilled occupational sectors under study in order for the applicants to be credible, and 3) there needs to be empirical evidence that this group has been discriminated against in order for the study to be relevant (for more details, see Larja et al., 2012, p. 144). Russians were chosen as the ethnic group for several reasons. Firstly, Russian-speaking immigrants constitute the largest immigrant group in Finland (23% of the total immigrant population at the end of 2013; Statistics Finland, 2014), making the results of a field study more socially relevant than those pertaining to smaller groups. Secondly, in order to study gender as a ground for recruitment discrimination, women from both the selected immigrant group and the national group should be commonly employed in various sectors. This is why some other large immigrant groups which have been shown to experience discrimination (e.g., the Somalians) were excluded. Thirdly, to our knowledge, discrimination against Russian job seekers in an EU country has not been studied before with the situation testing method.
In Finland, the Russians occupy an even lower position in the status hierarchy of ethnic groups than the Poles but share with them much of the general stereotype of Eastern Europeans (Jaakkola, 2005). However, unlike Poles who are a numerically small ethnic group, Russians-both male and female-are widely employed in many sectors (Statistics Finland, 2011), which enhances the credibility of the fictitious applicants. In addition, as was evident in the literature review above, the integration of the immigrants from the former Soviet Union into Finnish society has been far from smooth. Thus, it is hypothesised that Finnish-named applicants will be favoured over Russian-named applicants in recruitment in all job sectors (H1A).
Gender stereotypes of occupations affect the amount of discrimination against males and females (Carlsson & Rooth, 2008;Nunes & Seligman, 2000;. Thus, following previous studies on gender discrimination (e.g., Weichselbaumer, 2004) it was hypothesised that Finnish women would be discriminated against in stereotypically male and Finnish men in stereotypically female occupations (H2). As research on gender discrimination of immigrant job seekers is scarce and results inconclusive, we explored whether the validity of the pattern of gender discrimination in relation to occupational gender stereotypes hypothesised for the majority members holds true in the case of Russian-named male and female job seekers.

Method
The study design, sampling procedures, selection of vacancies, recruitment and training of research assistants, production of test materials, participant flow from one stage to another as well as validity tests on applicant homogeneity are described in full detail in a previous report (Larja et al., 2012). Main points are summarised here.

Situation Testing in the Field
In situation testing (see for example Bovenkerk, 1992), a research method where pairs of two applicants who differ only on the studied characteristic (e.g. ethnicity, gender) apply for the same jobs, the education and job history of the applicants are made as similar as possible without risking arousing the suspicion of recruiters. The role of applicants is played by research assistants (testers) who are carefully selected and trained to match in their conduct, attitude and appearance. A complete testing procedure consists of three stages: in the first inquiry stage testers call or visit the employer to ask whether the job is still available and if they can apply for it. In the application stage written applications (CV, cover letter) are sent to the employer, and the last stage is the job interview. Conducting field experiments is easier in the first and second stages as contact with the recruiters is re- Discrimination in Recruitment 412 stricted to either phone calls or written applications. In order to avoid excessive costs and many of the pitfalls of the third stage (e.g., matching of the applicants' appearance) (Pager, 2007), and because the largest share of discrimination on the ground of ethnicity has been found to occur in the second stage (Riach & Rich, 2002), we chose to conduct only the first two stages of the procedure.

Sample and Procedure
In Study 2, Russian-and Finnish-named female and male test applicants applied for a total of 1,258 vacant jobs at the end of 2011. Because of the limited number of vacant jobs, random sampling could not be applied. Instead, all the jobs were included that fit our criteria (see below). In the first inquiry stage our testers made altogether 1164 calls (= 582 valid tests, i.e., pairs of applicants with similar CVs but different ethnicity making an inquiry or sending in an application) to employers in response to an advertised job opportunity. If no phone number was provided, we moved directly to the application stage. In the first inquiry stage the critical outcome was an invitation to send in a written application. In the application stage altogether 1 690 written job applications (= 845 valid tests) were sent via e-mail. The critical outcome was the invitation to a job interview.
There were four possible outcomes in each stage: both applicants proceed in the process (i.e., are encouraged to send in their application or are invited to an interview), only the ethnic majority member, only the ethnic minority member, or neither of the applicants proceeds. Preferential treatment occurs in cases where only one applicant proceeds. Discrimination is considered to have occurred if one of the applicants is systematically invited to proceed in the recruitment process (e.g., to send in an application, is invited to an interview or offered a job) more often than the other one.
Choosing Occupational Groups -Study 2 was conducted in the main economic centres in Finland where most of the Russian-speakers live. The vacancies were among semi-skilled jobs, as recommended by Bovenkerk (1992, pp. 21-22, 27-29). In this study, testers applied for restaurant jobs (e.g., chef, cooker, waiter/waitress), semi-skilled office jobs (e.g., accountant, payroll clerk, secretary, receptionist), driver jobs (e.g., taxi driver, truck driver, lorry driver), and construction jobs (e.g., construction worker, painter, tiler). Semi-skilled office jobs are female-dominated (81% female employees), driver and construction jobs are male-dominated (93% and 96% male, respectively), whereas the restaurant sector is more ambiguous: Cook jobs are somewhat less female-dominated (71% female) than waiter/waitress jobs (77% female). Despite the relatively large amount of females in cook jobs, however, the professional stereotype for cooks in Finland is more masculine than feminine -perhaps due to celebrity cooks being largely male.
The vacancies were selected among positions advertised on the web pages of the Employment and Economic Development Office. Open vacancies were randomly assigned to the testers.
The Applicants and Applications -Research assistants playing the role of (fictitious) job applicants, four females and five males, were recruited among students of colleges and universities in Helsinki. Five of the testers spoke Russian and four spoke Finnish as their mother tongue but all of them could speak Finnish with a Russian accent.
They were trained for their tasks in a one-week preparation course by employers from the studied sectors, a professional actor and the researchers.
The applications consisted of a one-page CV and a cover letter, representing five different professional profiles.
All applications were written in standard Finnish with no language errors and the CV made clear that the Russian- named applicant was educated in Finland and fluent in Finnish. The ethnicity and gender of the tester were communicated primarily by the name and mother tongue of the applicant. During the phone inquiries, the ethnicity and gender of the applicant was communicated through name, voice and accent.
Acknowledging the criticism against situation testing (e.g., Heckman, 1998), at least two different testers (research assistants) played each of the four characters (male and female Russian and Finn). In this way, the results for one character are not dependent on only one person's performance. As suggested by Bovenkerk (1992, p. 39;Bovenkerk, Gras, & Ramsoedh, 1995, p. 12), also validity tests were conducted for the data and minor corrections were made accordingly (for more details, see Larja et al., 2012, p. 157).

Ethical Considerations -
The employers could not be informed of the study beforehand, as it probably would have altered the results. Although discrimination can be studied in laboratory experiments, only the results of field experiments can be generalised to the actual situation in the labour market. However, the absence of informed consent gives rise to ethical considerations. For this reason, the study design was presented to the Ethical Review Board in the Humanities and Social and Behavioural Sciences at the University of Helsinki and permission to conduct the study was granted.
In addition to matters of moral principle involved in misleading participants in the study, the employers inevitably lose when using their time to answer phone calls and process applications of people who are not genuinely interested in the open positions. Furthermore, the method entails a small risk that the success of real applicants is affected by the presence of fictitious applicants in the recruitment process. Thus, our choice to include only the first inquiry (phone call) and application stage and omit the job interview stage was motivated by the wish to minimise the inconvenience caused to employers and other job seekers. Furthermore, invitations to interviews were cancelled as soon as possible to minimise any inconvenience.
To safeguard the anonymity of employers, all persons involved in the study signed a non-disclosure agreement and it was ensured that employers could not be identified from the published results. The well-being of our research assistants (the testers) was monitored during the data collection. After the data collection, all employers were sent a debriefing letter with contact details of the researchers.

Ethnic Discrimination
As shown by the results (see Table 1), H1A predicting that Finnish-named (majority) applicants will be favoured over Russian-named (minority) applicants in recruitment in all job sectors was supported. In the first inquiry stage the call-back rate (CBR; the share of positive answers in response to phone calls) was 82% for the Finns and 77% for the Russians. As at this stage both applicants were paired due to applying for the same position, the  Note. For detailed results, relative call-back rates or net-discrimination rates, see Larja et al., 2012, pp. 165-166. M = masculine; F = feminine.
All tests are 2-tailed.
Results from the second stage were analysed also according to occupation and nature of the job. Due to small numbers of applications within these sectors, taxi, truck and other delivery van driver jobs were combined with construction worker jobs. In spite of the smaller subsample sizes in this analysis, the differences in call-back rates between applicants with Finnish and Russian names remained statistically significant in almost all occupational categories (see Table 1). Hence, discrimination occurred in all occupational categories.

Gender Discrimination of Finnish-Named Applicants
In order not to mix the effects of gender discrimination with the effects of intersectional discrimination (ethnicity and gender), the results for Russian-and Finnish-named applicants were analysed separately when addressing gender discrimination in recruitment for jobs characterised by different gender stereotypes. However, this caused sample sizes to drop considerably for most occupational sectors. This means that substantial differences in the call-back rates are required for the difference between the sectors to become statistically significant. Consequently, we combined the occupational sectors into two categories only, i.e., stereotypically feminine and stereotypically masculine, respectively.
In order to test H2 we first performed a logistic regression for the Finnish applicants to test the effects of gender, occupational gender stereotype and the interaction between these on the likelihood that the applicant is invited to a job interview. The logistic regression model was statistically significant (χ2(3) = 11.430, p = .010). The model explained 2.0% (Nagelkerke R 2 ) of the variance in getting invited to a job interview and correctly classified 74.2% of the cases. Regardless of occupational sector, gender was a significant predictor of being invited to an interview; women were preferred overall when recruiting members of the ethnic majority: Finnish women were 2.04 times (B = 0.714, p = .001) more likely to be invited into a job interview than Finnish men. The main effect of the occupational gender stereotype was not statistically significant.
However, and supporting our H2 on gender discrimination in recruitment being dependent on the occupational gender stereotype among Finnish majority applicants, the interaction between gender and the occupational gender stereotype was significant in the Finnish sample and thus predicted the likelihood to be invited into the job interview among Finnish job-seekers (B = 0.498, p =. 031). As can be seen in Table 2, female Finnish applicants were invited to a job interview more often than Finnish males when applying for stereotypically feminine jobs, i.e., male applicants with Finnish names were discriminated against in stereotypically feminine jobs (office and waiter/waitress jobs).
The CBR was 19% for men and 32% for women, which suggests that male applicants from the ethnic majority Liebkind, Larja,& Brylka 415 group needed to submit almost twice as many applications as a female applicant from the same ethnic group in order to obtain to an interview for an office job.
However, there was no statistically significant discrimination of Finnish women in stereotypically masculine sectors (drivers, construction workers and cooks), and H2 on ethnic majority women being discriminated against in stereotypically masculine sectors was not supported. Therefore, only partial support was found for H2 on gender discrimination of Finnish-named applicants in stereotypically feminine and masculine jobs, as only men were discriminated against in the stereotypically feminine sectors.

Gender Discrimination of Russian-Named Applicants
To test our second research question, i.e., whether or not gender discrimination in recruitment is dependent on occupational gender stereotypes in the same way among majority and minority applicants, and thus to explore whether H2 would apply also to the minority members, i.e., whether the Russian-named ethnic minority applicants faced similar gender discrimination as Finnish-named majority applicants in occupations characterised by different gender stereotypes, the analysis of gender discrimination was repeated with only Russian-named minority applicants. Note. For detailed results, relative call-back rates or net-discrimination rates, see Larja et al., 2012, pp. 172-174. All tests are 2-tailed.
Interestingly, the results did not follow the same pattern as with the Finnish-named majority applicants. At first, the results look similar to those obtained for the Finnish sample. The logistic regression model conducted for the Russian sample to test the effects of gender, occupational gender stereotype and the interaction between these was significant (χ2 (3) = 10.020, p = .018). The model explained 2.2% (Nagelkerke R 2 ) of the variance in getting invited to a job interview and correctly classified 87.1% of cases. As in the case of the Finnish sample, the main effect of the occupational gender stereotype was not statistically significant but that of gender was, so that Russian women were 1.99 times (B = 0.693, p = .015) more likely to be invited into a job interview than Russian men regardless of occupational sector. In contrast to the results obtained for the Finnish sample, however, the interaction between gender and the gender stereotype of the occupation was not a significant predictor for the likelihood of Russian applicants to be invited into a job interview. The results from the chi-square tests (cf. Table 2) show that ethnic minority women were not treated worse than ethnic minority men in the stereotypically masculine sectors of driving, construction and cook jobs. Surprisingly, even an opposite tendency was observed, i.e., that Russian women tend to be treated better than Russian men in typically masculine job sectors (see Table 2). With regard to stereotypically feminine jobs (office and waiter/waitress jobs), the discrimination of Russian men compared to Russian women is significant, as was the case of Finnish men compared to Finnish women in stereotypically Discrimination in Recruitment feminine jobs (see Table 2). Therefore, the pattern of gender discrimination among the ethnic minority applicants did not follow the pattern hypothesised for the ethnic majority applicants (H2). Specifically, ethnic minority men were treated significantly worse than ethnic minority women in both stereotypically feminine (office clerks and waiters/waitresses), and masculine job sectors (driving, construction and cook), while majority men were treated significantly worse than majority women only in the stereotypically feminine occupations. Among both minority and majority applicants, women were not discriminated in stereotypically masculine jobs.

General Discussion
As regards our first research question on the relationship between recruitment discrimination and the position of the applicants' ethnic group in the local ethnic hierarchy, the results of Study 1 clearly showed that a foreign applicant from a low-status ethnic minority group (Polish) had significantly lower chances of being selected for a vacant position when paired with a majority applicant (Finnish), as well as when paired with an applicant from a high-status ethnic minority (Austrian). However, as the high-status minority applicant was not discriminated against when paired with the Finnish applicant, our Hypothesis 1A was only partly supported by Study 1: Ethnic minority membership by itself does not seem to be a sufficient ground for ethnic discrimination in recruitment. Instead, a sufficiently low position in the local hierarchy of the applicant's ethnic group seems to be the ground for discrimination. This claim is further supported by the finding that not only the majority applicant but also the Austrian applicant was preferred over the Polish applicant, thus confirming our Hypothesis 1B. Consequently, it is not the applicant's group membership as such but rather the rank of the applicant's group membership in the local consensual ethnic hierarchy that seems to be more informative about the pattern of recruitment discrimination that applicants are likely to face due to their ethnicity (Jaakkola, 2005(Jaakkola, , 2009).
It should be noted that the position of one's membership group in the ethnic ranking affected the amount of recruitment discrimination even though the recognition of the professional qualifications of the applicants was held constant. To our knowledge, this study is the first one to show that even if immigrant applicants have unified and equally recognised qualifications, the factor which is decisive for their employment in a position matching their professional qualifications in the host country is their nationality due to its associated rank in the ethnic hierarchy.
The results of Study 2 widened the occupational domains studied in Study 1 from skilled to semi-skilled jobs and gave further support to Hypothesis 1A on the effect of ethnic ranking on recruitment discrimination. Specifically, the results corroborated that a Russian applicant had significantly lower chances of getting interviewed for a vacant position in the studied occupational sectors. Regardless of the type of occupation, in the second stage of Study 2, 13% of the Russian and 26% of the Finnish applicants were invited to a job interview. This implies that the RCR (relative call-back rate) equals 2.00, which means that compared to the Finnish applicant, the Russian applicant needed to send twice the amount of applications to receive as many interview invitations.
Although the results of Study 2 cannot fully represent general national levels of discrimination in Finland and although many factors can influence RCR-levels in different studies, some cautious cross-national comparisons can be made. The obtained RCR of 2.0 is somewhat higher than the average RCR in other situation tests carried out in the 2000s in Europe (1.8) and in Canada, the United States and Australia (1.6) (Larja et al., 2012, pp. 131-134). This result may be indicative of the differences in the integration of immigrants and attitudes towards immigrants in traditional nation states versus so called settler states with a longer tradition of immigration; in the latter Liebkind, Larja,& Brylka 417 countries, attitudes towards immigrants are usually less negative (see, for example, Phinney, Berry, Vedder, & Liebkind, 2006;Ward & Masgoret, 2008). In this respect, Finland is a nation state which belongs to more recent immigration destinations and this is reflected in the discrimination experiences of ethnic minorities (e.g., FRA, 2009, p. 36) being more numerous and in the attitudes of the majority population towards ethnic minorities (e.g., Eurobarometer, 2008) being slightly more negative than in other EU countries which have a longer history of immigration.
The RCR of 1.08 obtained for the first (call back) stage shows that Russian applicants were discriminated against when compared to Finnish applicants already at this initial point of the recruitment process. This RCR is similar to the one found in Sweden for Middle Eastern applicants (1.06; Attström, 2007, our calculations) and although it already points at some discrimination against minority applicants, it also underlines the point made by Attström (2007, p. 44) that substantial and selective screening out of applicants occurs only later when employers choose whom to call and invite to an interview. Obviously, in both Finland and Sweden, political correctness when minority applicants call (i.e., unwillingness to explicitly discourage minority members to apply even when the employer has no interest in the applicant) has resulted in comparably lower discrimination rates in the first stage of the recruitment process as compared to the discrimination in the second stage.
Our second research question was whether or not gender discrimination in recruitment depends on occupational gender stereotypes in the same way among majority and minority applicants. We assumed (H2) that Finnish men will be discriminated against in recruitment to occupations considered to be feminine, while Finnish women will be discriminated when the occupations are considered to be masculine. In Study 2, our H2 was supported in that we observed a significant interaction between gender and the occupational gender stereotype in the majority sample. However, a closer look revealed that H2 was only partly supported as, among the Finnish majority group, male applicants were found to be discriminated against when applying for stereotypically feminine jobs, but this was not the case for Finnish women applying for stereotypically masculine jobs. These findings were thus only partly in line with previous results from situation tests in other countries. Although some previous results have shown discrimination against women in typically 'masculine' occupations (Nunes & Seligman, 2000;Riach & Rich 2006; but for an exception see Carlsson & Rooth, 2008), such effects were not found in this study. Thus, the results of Study 2 clearly differ from those obtained by Arai et al. (2011) who found no gender discrimination among the native Swedish applicants in the normal-CV-setup and strong evidence of discrimination against Swedish women compared to Swedish men in almost all occupational categories when immigrant applicants had better CVs than native Swedes. However, situation tests for the most part use either entry-level or semi-skilled jobs and women are typically discriminated when applying for high status posts (e.g., Riach & Rich, 2002;Petit et al., 2011). Therefore, this particular research design may not be fully suitable for studying gender discrimination, at least not in the Nordic countries where gender equality is prioritised (e.g., Bergqvist, 1999). Another explanation for this result may be found in studies on wage discrimination (Asplund, 2008;Nieminen, 2008): The problems of women in Finnish working life may reside more in wage discrimination and promotion opportunities than in recruitment discrimination.
When the effects of gender and occupational gender stereotypes were explored among the Russian applicants in order to address our second research question, the pattern seemed to be different: The results suggest that in the case of these minority applicants, gender discrimination cannot be predicted by the stereotypical "masculinity" or "femininity" of the occupation; the interaction between gender and the occupational gender stereotype was non-significant among the minority applicants. Partly consistent with previous findings of Arai and colleagues  (2008,2011) on job seekers with Arabic names, men with Russian names had significantly lower call-back rates than Russian-named women in stereotypically feminine (office and waiter/waitress) jobs but, surprisingly, tended to have lower call-back rates than Russian-named women also in masculine (construction/driver and cook) jobs.
However, females with Russian names were not discriminated against on the ground of gender, although they were still discriminated against on the basis of their ethnicity in all occupational sectors when compared to Finnish female applicants. These results seem to corroborate the conclusions drawn by Ghavami and Peplau (2013) that ethnic minority women may be stereotyped very differently from the stereotypes of their ethnic group, and that culturally held intersecting gender and ethnic stereotypes contain unique elements that cannot be obtained by simply adding gender stereotypes to ethnic stereotypes. Arai et al. (2011) who studied workers from Arabic countries argued that working women in general are perceived as less traditional and that Arabic women are thus less likely than Arabic men to be judged through the national stereotype of Arabs. The situation of working Russian women is, however, different. Although the position of employed women in Russia declined somewhat after the fall of the Soviet Union (from 51 percent in 1990 to 48 percent in 1998), the position of women in Russia as relatively equal partners in the labour market still prevails (Rzhanitsyna, 2000). Thus, simply being in the labour market cannot 'protect' working Russian women from being judged through the national stereotype of Russians. However, in the case of Finland, the historically-grounded negative stereotypes of Russians tend to reflect a perceived threat (Stephan & Renfro, 2002) of (military or other) aggression (Karemaa, 1998). These negative stereotypes of Russians in general could be assumed to apply more to Russian men than Russian women, who thus would be perceived as less threatening and more likeable than Russian men, regardless of the context. This could possibly explain why we found even in the stereotypically masculine job sector a tendency of Russian women to be treated better than Russian men. Therefore, more research with Russian-named male and female applicants in other countries needs to be conducted, duly acknowledging the ranking of this ethnic group in the local ethnic hierarchy.
In summary, the studies reported here contribute to existing knowledge on recruitment discrimination by (1) using the situation test method in a context (Finland) where it has not been used before; (2) studying the recruitment discrimination of a group (Russian immigrants) which has not been studied before in any EU-country; (3) covering a more diverse domain for discrimination in recruitment compared to previous studies from skilled (Study 1) to semi-skilled jobs (Study 2); (4) studying, for the first time, recruitment discrimination on the ground of ethnicity towards low status ethnic minority groups from within and outside EU while controlling simultaneously for the possible confound of formal recognition of qualifications (in addition to skills and work experience), and by (5) addressing the rarely studied issue of gender discrimination in relation to occupational gender stereotypes in the recruitment discrimination of immigrants.
Besides the novel contributions which our research makes to the existing literature on labour market discrimination, also the limitations of the present studies must be acknowledged. would be expected to face less discrimination. It is also possible that applicants with Somali-sounding names would have faced even more discrimination than did the Russian-named applicants as their position in the ethnic hierarchy is even lower (Jaakkola, 2005(Jaakkola, , 2009).
The generalisability of the results could be to some extent affected also by the selection of the occupational categories, cities and period of data collection. However, although our sample of vacancies was not random, it included all vacant positions in Finland that fit our criteria among the largest occupation categories (i.e., by the number of open vacancies) during the data-gathering period. The results of this study are thus likely to largely represent the studied occupational categories at the given time. However, as only around 40% of vacancies in Finland are announced publicly (Tuomaala, 2009, p. 9) and the field experiment covered only the main population centres in Finland (28% of the total population; Statistics Finland, 2013b), the representativeness of our job sample can still be considered limited. In addition, the results of this study cannot automatically be generalised to other occupational sectors (see also Bursell, 2007) and especially gender discrimination should in future research be studied in more high status professions.
As Study 2 did not encompass the third stage of the recruitment process which is the job interview, the final outcome of the recruitment process, i.e., the employment decision, remained unknown. However, as only a small part of the discrimination occurring during the entire recruitment process takes place in the third stage (Larja et al., 2012, p. 135), it is reasonable to assume that this study covered the largest share of discrimination cases. It is also worth noting that Finland, along with the rest of Europe, was undergoing an economic recession at the time of data gathering. Thus, it is plausible that the results may be different during a period marked by more economic optimism and better employment prospects. However, as the recession in Finland has only deepened since Study 2 was conducted (Ministry of Finance, 2014), concerted action against recruitment discrimination is more urgent than ever.
Considering this urgent task, the practical relevance of the results obtained in the two studies is obvious: They can be used to raise the awareness of employers of their own recruitment behaviour. In order to create a fairer labour market, such awareness-raising is necessary, especially as employers may discriminate unknowingly.
Evidence of such unconscious bias was found in a study by Rooth (2010) who found that the implicit (automatic, subconscious) attitudes of Swedish employers whose implicit attitudes had been tested prior to their recruitment behaviour were associated with a reduced probability to invite minority members to a job interview, while explicitly stated attitudes did not predict employers' discriminatory behaviour. As noted in the literature review above, recruitment discrimination represents an unjustifiable negative behaviour towards a group or its members on the basis of their group membership, regardless of underlying motivations or the degree of conscious intent (Al Ramiah et al., 2010). Future research should further specify the psychological processes responsible for such behaviour. We can only speculate that, in accordance with some previous research results (Agerström & Rooth, 2009Rooth, 2010) implicit rather than explicit (deliberate) attitudes are responsible for the results we obtained on recruitment discrimination in this study.

Discrimination in Recruitment 420
Recruitment discrimination on the basis of ethnicity or gender is against the law, on national as well as EU levels.
Yet the results from Study 1 indicate that although EU citizens from all member states have the same rights within the EU, citizens from newer member states may be discriminated against in a recruitment situation. Such discrimination seems to occur not only when applicants from the newer EU member states are compared to equally qualified local applicants but also to equally qualified applicants from older EU states, even when controlling for the formal recognition of their professional qualifications. Results from Study 2, in turn, revealed that illegal ethnic discrimination in recruitment actually takes place on a relatively large scale. Clearly, the problem is not unique to Finland, but this is the first time it has been reliably identified. Although anonymisation of applications has been offered as one solution to this problem (e.g., Carpusor & Loges, 2006), it seems insufficient for the recruitment situations tested in this study. More specifically, anonymisation is relevant only with later generations of immigrants who no longer reveal their foreign background with their accent and/or origin of their qualifications. Thus, the findings of the present research point at the importance of alerting employers to the fact that they may, even unwillingly, break the law in the recruitment process.  [1939][1940]. The Winter War, being a military conflict between the superpower and the nation of around 3,5 million inhabitants, remains for Finns a symbol of heroism (Finell, 2012).

Notes
iii) Additional analysis. As previous research in Finland (e.g., Jaakkola, 2005) has shown that women have more positive attitudes towards immigrants than men, an additional chi-square test was performed to assess whether the obtained result differed for male and female participants. Due to the small percentage of male participants in Conditions 1 and 3, the chi-square test could not be performed for each of the conditions separately, as the minimal requirements for performing such a test were not met (see Bryman & Cramer, 2005). However, we combined the overall choice of the Finnish, Polish and Austrian applicant across conditions, which allowed for testing the effect of gender in the whole sample. The results showed that the preference for employing the applicants was indeed gender-dependent χ 2 (2) = 7.551, p = .023. A closer inspection of the results (actual versus expected count) revealed that the Austrian applicant was chosen by female participants more often and by male participants less often than expected, and, conversely, that the Finnish applicant was chosen by female participants less often and by male participants more often than expected. However, there was no gender difference in recruiting the Polish applicant, as the actual and expected counts were the same.

Funding
This research project was co-financed by the Ministry of Employment and Economy and the Ministry of Interior (Finland) and the YES 4 -Equality is Priority project, supported by the European Union PROGRESS Program (2007)(2008)(2009)(2010)(2011)(2012)(2013).