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

student populations. Lottery-based admission will therefore still result in an overrepresenta- tion of high SES students without a migration background, and will not ensure the entry of low SES applicants with an underrepresented migration background [17]. Furthermore, an interview study with Dutch applicants to undergraduate health professions education pro- grams shows that there is no (sub)group of applicants which favors lottery, regardless of whether this is a weighted or a random lottery. They prefer selection instruments on which they can feel more ‘in control’ [64]. The data in this study suggests that in the Dutch context, equitable admissions procedures (which give more opportunities to applicants from underrep- resented groups which face significant barriers on the road to HE), have a better chance of increasing the (intersectional) diversity of student populations in selective HE programs than a lottery can achieve. These findings can be relevant in other higher education contexts where applicant pools are also not representative of the general society. We recommend improved national monitoring in the future of the (potential) inequality of opportunity in HE selection procedures, and its effects on the (intersectional) representative- ness of the student population. Furthermore, as Steenman argues, “using selection is generally seen as limiting the standard idea of accessibility and should therefore be used sparingly” [65]. HEIs could consider whether, in the absence of government-mandated capacity limitations, it

is justified to use a selection procedure as long as there is no guarantee that this does not harm equality of opportunity and student diversity. Furthermore, we recommend a detailed study of the selection procedures in the only cluster which did not show any inequality of opportunity, as other selective programs may be able to learn from the Artificial Intelligence and Industrial Design selection procedures in the year 2020. In the same way, study programs in which certain groups are underrepresented (e.g. on the basis of sex or migration background) and/or had lower odds of admission, may be able to learn from other clusters where these underrepresented groups had an advantage in the selec- tion procedure. On methodology. Our findings show that it is important to study inequality of opportu- nity through an intersectional lens [8]. The acceptance rates of groups ALL versus NONE indi- cate the vast differences of odds of admission between applicants who either have all the beneficial ‘check marks’ [60] in a selection procedure, and those who have none. We therefore recommend future quantitative (international) research to build on our methodology, when- ever suitable, to create intersectional analyses based on the multiplicity of identity layers that people have. We believe our method to be easier to interpret than traditional regression mod- els with interaction terms. This is especially relevant when the research focuses on inequalities between groups. In order to address and resolve inequities, such as in the educational sphere,

it is important that these inequities are clearly demonstrated in a way that all audiences can understand. This is what Table 3 in this article has intended to do. Limitations A limitation is that we could not use parental education levels, as these data are largely missing for parents who were educated outside of the Netherlands. For parents who were educated in The Netherlands, educational achievement is not sufficiently standardized. We circumvented this limitation by including other SES variables that served as proxies. In the original data, assets percentiles are registered at the household level, on the person who is the main breadwinner. When parents are divorced and live in separate households with partners who each are the main breadwinners, this results in missing data on parental assets. For a few clusters, parental income and assets data was missing for more than 20% of appli- cants (namely in SU3, SU7, SU8 and SU9). It is possible that in these cases, applicants were e.g. 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 19 / 24