After the wave of studies completed in the 1990s, the topic of discrimination received renewed attention from academics. After 2000 the number of studies in the US and Europe increased quickly. Particularly in Europe, where the European Union had adopted anti-discrimination directives which EU member states had to transpose into national law, correspondence tests were carried out in many EU countries, covering almost all countries of Western Europe. In the 2000s the last three ILO studies on discrimination in the labour market were published for Italy (Allasino et al., 2006), France (Cediey & Foroni, 2008) and Sweden (Attström, 2007). Once the Swedish ethical approval to join the ILO project was granted, numerous correspondence tests were conducted in Sweden, Norway, and Finland, where researchers had previously assumed that their research would not be approved either (Larja et al., 2012; Liebkind, Larja, & Brylka, 2016; Midtbøen, 2013, p. 52). Next to the Scandinavian countries, studies were, amongst others, implemented in Austria
(Weichselbaumer, 2016b), Belgium (e.g. Baert, Cockx, Gheyle, & Vandamme, 2015), France (Duguet, Leandri, L'Horty, & Petit, 2010), Germany (Kaas & Manger, 2012; Schneider et al., 2014), Greece (Drydakis & Vlassis, 2010), Ireland (McGinnity & Lunn, 2011), the Netherlands (Andriessen, Nievers, Dagevos, & Faulk, 2012), Switzerland (Fibbi, Lerch, & Wanner, 2006), or
the UK (Wood, Hales, Purdon, Sejersen, & Hayllar, 2009). Despite the strong focus on North America, Europe and Australia, the methodology of testing for ethnic or racial discrimination was also employed in countries such as Chile (Bravo, Sanhueza, & Urzúa, 2008), China (Maurer-Fazio, 2012), Georgia (Asali, Pignatti, & Skhirtladze, 2017), India (Banerjee, Bertrand, Datta, &
Mullainathan, 2009), Israel (Ariel et al., 2015), Malaysia (Lee & Khalid, 2016), or Mexico (Arceo- Gomez & Campos-Vazquez, 2014).
Field experiments have also addressed specific segments of the labour market. While Pager et al. (2009) focused on the low wage labour market, Bendick, Rodriguez, and Jayaraman (2010) tested for racial discrimination of waiters in up-scale restaurants. Two of the correspondence tests conducted in Germany also focus on very specific labour markets, i.e. the market for
apprenticeships (Schneider et al., 2014) and the market for student internships (Kaas & Manger, 2012). University graduates have increasingly become the focus of correspondence studies, as can be observed in the US with studies conducted by Nunley, Pugh, Romero, and Seals (2015), Gaddis (2014), or Deming, Yuchtman, Abulafi, Goldin, and Katz (2016).
Moreover, field experiments have evolved from the original paired within subject design which varies only the dimension of race or ethnicity. Oreopoulos (2011) conducted one of the most comprehensive and complex field experiment on ethnic discrimination with almost 13,000 resumes on the Toronto labour market. By differentiating the fictitious CVs by name, place of education and place of work experience, he was able to show that Canadian employers discriminate by name and that Canadian work experience was more important than Canadian education. This experiment was repeated by Dechief and Oreopoulos (2012) with a greater regional scope, but overall similar results. Furthermore, field experiments no longer focused on only one minority group, but started to compare different ethnic or racial groups (e.g. Wood et al., 2009), and even added indigenous populations (Booth, Leigh, & Varganova, 2012).
Furthermore, research designs have become increasingly complex trying to go beyond showing the mere existence of discrimination in the hiring process, but attempting to narrow down underlying reasons. New variables that have been added next to race or ethnicity include further information included the CV, studying the intersection of gender and/or religion, or focusing on the context of the application. Specifically, researchers tested the impact of:
- varying the resume quality or residential area of an applicant (Bertrand & Mullainathan, 2004; Jacquemet & Yannelis, 2012),
- varying the resume quality, language skills and quality of the university (Lee & Khalid, 2016), - indicating a criminal record on the CV (Pager, 2003),
- whether university degrees were obtained at an elite or less selective university (Gaddis, 2014), or from a public or for profit institution (Deming et al., 2016),
- including a reference letter from a previous employer (Kaas & Manger, 2012),
- including productivity variables for college graduates e.g. information on unemployment periods or internships (Nunley, Pugh, Romero, & Seals, 2014),
- signalling candidates status as first generation immigrant or second generation youth (M. Carlsson, 2010),
- adding foreign names8 whose origins were not easily recognised by Americans (Jacquemet &
Yannelis, 2012),
- the intersection of race and gender (Darolia, Koedel, Martorell, Wilson, & Perez-Arce, 2016), - whether applicants wore a head-scarf (in the US Ghumman and Ryan (2013), in Germany
Weichselbaumer (2016a)),
- CV whitening (Kang, DeCelles, Tilcsik, & Jun, 2016),
- Facebook profile pictures (Baert, 2015), indicating work experience (Baert, Albanese, du Gardein, Ovaere, & Stappers, 2017), or volunteering (Baert & Vujić, 2016)
- whether employers were in the public or private sector (e.g. Wood et al., 2009),
- submitting applications as CV or if application forms were used e.g. Wood et al. (2009), - the gender of the contact person in the vacancy (M. Carlsson, 2010), or
- whether labour markets were tight or not (Baert et al., 2015).
Drydakis and Vlassis (2010) also complemented their Greek correspondence test with information on wages and insurance coverage offered. Blommaert, Coenders, and van Tubergen (2014)
conducted a field experiment on hiring discrimination in the Netherlands, in which they posted resumes on online job search websites and measured response rates from employers, by counting the number of times the profile was viewed as well as the times candidates were contacted by potential employers. While this is not a classical correspondence test, the reported results of discrimination were rather similar.
8 Names were constructed by using Albanian, Armenian and Georgian first names, groups which are relatively unknown in the area of Chicago. These first names were combined with male last names from the same groups. Reasoning that these names cannot be attributed to a specific ethnic group, Jacquement and Yannelis expected them to be just categorised as foreign and unfamiliar by employers (Jacquemet & Yannelis, 2012, p. 826).
As well as the added variables in the research design, mixed-methods and interdisciplinary
approaches have gained importance in the last years. Midtbøen (2014) complements his Norwegian correspondence test with employer interviews, thus combining different research designs. In a more interdisciplinary approach that leans strongly on the discipline of psychology, other researchers focus on unconsciously working stereotypes, implicit attitudes and automatically activated associations focusing mainly on recruiters to study the reasons of discriminatory behaviour in depths. Examples of these studies include Bertrand, Chugh, and Mullainathan (2005) who study implicit prejudices and behaviour, Agerström and Rooth (2009) who use Implicit Association Tests (IAT) to test whether employers in Sweden discriminate automatically against Arab Muslim
applicants, or Rooth (2010) who also focuses on Swedish recruiters and automatically activated associations that influence their behaviour and cause discriminatory treatment. However, more recent work on IATs has been more sceptical regarding the IAT’s ability to predict discriminaton (R. Carlsson & Agerström, 2016; Frederick L Oswald, Mitchell, Blanton, Jaccard, & Tetlock, 2013; Frederick L.; Oswald, Mitchell, Blanton, Jaccard, & Tetlock, 2015)
Recently research on the consequences of discriminatory behaviour on the employers has also emerged. Pager (2016) was able to show that employers who had been found to discriminate in her field experiment (Pager et al., 2009) were more likely to be out of business six years later. French research (Bon-Maury et al., 2016) also looked at the potential costs of discrimination to employers.