Inequality of opportunity in selection procedures limits diversity in higher education: An intersectional study of Dutch selective higher education programs

that applicants who were female, were born in 2001 or earlier, had parents in the income cate- gory 71–90, had 1 or 2 healthcare professional parents, had no migration background, and who lived in a postal code area which was averagely to weakly/not urban, had significantly higher odds of admission. Cluster SH1_2020 had 25 applicants in total who met all six criteria, and 20 of them received an offer of admission. This means group ALL had an acceptance rate of 80%. On the other hand, group NONE had an acceptance rate of 15.4%. As the number of persons accepted in group NONE was smaller than 10, Statistics Netherlands regulations do not allow the publication of the precise frequencies. However, the ratio of 80/15.4 = 5.19 indi- cates that applicants in group ALL were offered a place more than 5 times as often as applicants in group NONE. The ratios in Table 3 show that depending on the cluster, group ALL was admitted between 1.06 and 9.01 times as often as group NONE. The study clusters with the highest number of sig- nificant variables in the multivariable models are HBO clusters, resulting in the largest differ- ence in acceptance rates between group ALL and group NONE (see Ratio column in Table 3). To advance our intersectional understanding of (dis)advantage, Fig 2 gives the example of cluster SU4_2020. Two variables were significantly associated with odds of admission: migra- tion background and parental assets category. Acceptance rates differed vastly between groups: 50.1% for applicants without a migration background, 36.4% for TMSDI applicants, and

43.4% for applicants with an other migration background. When we divide these three groups based on parental assets category, we see additional discrepancies. Acceptance rates differed the most between applicants who had a TMSDI migration background and whose parents belonged to the lower assets category (28.8%) versus applicants who had no migration back- ground and whose parents belonged to the higher assets category (50.9%). Those with a TMSDI migration background and parents in the higher assets category, had a lower accep- tance rate (37.3%) than the applicants without a migration background or with a different migration background from the lower assets category (38.7% and 39.8%). In S1 File, similar figures can be found for all clusters. These illustrate how the variables which were significant in the multivariable model, result in (vastly) different acceptance rates for different (intersec- tional) groups, which diverge from the average acceptance rates as included in Table 1. The intersectional acceptance rates in S1 File add value to our understanding of inequality of opportunity in the selection procedures for higher education. They show that different groups of applicants (based on a combination of e.g. sex, migration background, and other variables) have very different chances of success in different programs. For example, in cluster SH5_2020 (Creative Media and Game Technologies; Fashion & Textile Technologies; Indus- trial Design Engineering; and Art and Economics), female applicants from (very) strongly

urban areas without a migration background had an acceptance rate of 80.8% (198/245). In comparison, the same cluster only accepted 56.0% of female applicants from (very) strongly urban areas with a TMSDI migration background (28/50). In cluster SU1_2019 (Medicine), female applicants without a migration background had an acceptance rate of 48.4% (1504/ 3107), while this was only 32.0% for male applicants with a TMSDI migration background (81/253). The latter percentage is significantly lower than the average acceptance rate of the cluster: 45.4%. Combined with the fact that male TMSDI applicants comprised the smallest group of the applicant pool (with only 253 applicants out of 6571 applicants - 3.9%), their low odds of admission contribute to their underrepresentation in the student population. This is part of a pattern found in many clusters. The result is that the student population of many selective programs does not reflect the diversity of the age cohort, the higher general or the pre-university population (see S3 Table). PLOS ONE Inequality of opportunity in Dutch selection procedures limits diversity in higher education PLOS ONE | https://doi.org/10.1371/journal.pone.0292805 October 13, 2023 15 / 24