A fair score? : group independent validity arguments for college admission tests
Abstract: This thesis aims to investigate fair use of test scores for admission to higher education. In this thesis, fairness is narrowed down to an instrumental definition relating to a fair interpretation of test scores. The focus is on the use of the Swedish Scholastic Aptitude Test (SweSAT) for admission decisions. The main questions targeted are whether the scores generated from an adaptation with extended time for test takers with dyslexia are fair and whether the SweSAT-scores may disadvantage students with an immigrant background.The thesis is based on validity theory and an established argument-based validity concept is used to link the results to crucial aspects of fairness. This thesis adds explicit fairness-related issues in some detail to this already established framework used for validity research related to the SweSAT. This approach involves some propositions about fairness, each of which can be investigated empirically which, to some extent, has been conducted in separate studies in this thesis.The individual studies contain a variety of results. In the first study, sub-scores of the SweSAT was found to have differential values for certain groups of test takers. This could cause unfairness if the sub-scores are used for admission decisions, but it also indicates that the internal structure of the test is inconsistent across groups. In the second article, the SweSAT was found to be speeded, meaning that the time limit is affecting the test score, and that test takers with dyslexia generally benefit from being given extra time to complete the test. It could not be ruled out that the extra time causes a beneficial experience of lower speededness compared to the rest of the test population, who take the test with no extra time. In the third article, we discover, to our surprise, that immigrant groups generally perform worse than expected in higher education when the expectations are based on their SweSAT and upper secondary school grades. In other words, the instruments used does not create an unintentional obstacle for immigrants in general since the ability of these groups to complete higher education is overestimated when grades and SweSAT-Scores are used for selection. However, in the final article, test items that may cause an unfair disadvantage for immigrant students was identified.In summary, the results contribute to expanding and, to some extent, strengthening the validity of using the SweSAT for admission. However, certain fairness issues have been noted.
CLICK HERE TO DOWNLOAD THE WHOLE DISSERTATION. (in PDF format)