Evidence-informed master admission

Abstract

This dissertation has the purpose to contribute to the science and practice of admission procedures in higher education. A literature analysis showed that the student selection literature is dominated by studies into medical and notably bachelor programs. Studies into master programs and non-medical programs, especially non-medical master programs, are vastly underrepresented. Furthermore, within existing medical and bachelor studies there is a lack of knowledge on the predictive validity of two often-used instruments: cognitive relatedness of prior education and application essays. Finally, there is gap of knowledge on how these instruments influence the diversity of the admitted cohorts of master students. This led to the following research question: How do cognitive and non-cognitive admission instruments influence the study success of individual master students and the diversity of groups of master students? This is done by the following 4 important contributions: 1. Non-cognitive admission instruments are an important aspect of admission procedures, because they are essential in accurately predicting the students’ necessary skills. Furthermore, there is a lack of information on non-cognitive admission instruments in the student admission literature. 2. Bachelor grades cannot be used as an admission instrument without caution because their predictive validity interacts with the relatedness in knowledge between bachelor and master programs. For students with weakly related bachelors, bachelor grades are only half as effective as an admission instrument compared to students with strongly related bachelors. 3. Application essays can certainly be an effective admission instrument and should not be dismissed. However, an important prerequisite is that the rating of such letters is structured and unbiased. I applied a text mining algorithm (Latent Dirichlet Allocation) to application essays and showed that five of seven unique topics in application essays correlate significantly with study success. 4. Bachelor grades, relatedness of prior education and application essays can be used as admission instruments without consistent negative effects on diversity of the student cohort. Contrary to belief and other studies, these admission instruments often have no negative effect on diversity. In fact, the effect is sometimes positive. And if the effect is negative, this only happens beyond certain admission thresholds. These contributions were formulated based on a variety of research methods. We used a systematic literature analysis incorporating human resource literature to give insight into non-cognitive admission instruments. Then, we introduced novel quantitative methods to this academic field, in the form of cosine similarity measurements, Latent Dirichlet Allocation and simulation models.