NEWS Are certain groups favored or disadvantaged by the design of the test? This is what a new dissertation from Umeå University has investigated.
ImageMattias Pettersson
In his dissertation, Marcus Strömbäck Hjärne has examined various aspects of whether the test is a fair selection instrument.
In the dissertation, I argue that the test as a selection instrument for higher education depends on the selection process being fair when comparing different groups. More specifically, I have studied the adaptation with extended writing time offered to dyslexics and also whether the test potentially disadvantages students with an immigrant background. The reason I looked more closely at the adaptations for dyslexics is that this group has grown very quickly and there have been discussions about offering extended writing time to more groups of students with various disabilities, despite the current adaptation being relatively unexplored. The admission of students with an immigrant background is also a relatively unknown area. We know from previous research that this group generally performs worse than native Swedes with Swedish parents, both in gymnasium and higher education, as well as on the SweSAT. The test has a relatively large verbal component where proficiency in Swedish plays a crucial role, while both gymnasium and higher education may compensate for certain language deficiencies. Therefore, I wanted to investigate whether a potentially irrelevant linguistic component in the test could be unfair to students with an immigrant background and others with language difficulties, says Marcus Strömbäck Hjärne.
Marcus Strömbäck Hjärne, Department of Applied Educational Science. Image: Private
In his analyses, Marcus Strömbäck Hjärne can also see that there are parts of the test that can cause unfair differences between groups.
My results show, among other things, that dyslexics generally benefit from getting extra time during the test, but I cannot rule out that the extra time offered to dyslexics leads to them experiencing less time pressure compared to everyone else taking the test without extra time. I can also show that the test is time pressured. This is even though speed is not a trait intended to be measured. The result can be problematic from a fairness perspective because the goal of the extra time is for dyslexics to take the test under the same conditions as other participants.
In one of the studies, I also discovered, to my surprise, that students with foreign-born parents generally perform worse than expected in higher education. In that study, the proportion of completed credits during the first year is used as a measure of academic success, and the expected result is based on regression models with grades and test results as predictors. Since age, gender, parents' education level, and the type of university education the student is studying have an impact on higher education, these factors have been controlled for. The results that students with an immigrant background, all else being equal, generally perform worse in higher education compared to students whose parents are born in Sweden indicate that there may be something in higher education that disadvantages university and college students with an immigrant background, continues Marcus Strömbäck Hjärne.
In summary, Marcus Strömbäck Hjärne's dissertation shows that there are systematic group differences in the selection for higher education, which means that certain groups are underrepresented in higher education. However, based on the analyses that the dissertation has focused on, the differences in outcomes cannot be explained by the test itself unfairly favoring or disadvantaging specific groups. He can also see that there are parts of the test that need further research to make it as fair as possible.
For example, we need to know more about the possibility of offering other types of adaptations or reducing time pressure for everyone. We also need to continue studying immigrant groups in more detail to address the causes of the selection's overestimation of this group. In the long run, the results of the dissertation, if applied correctly, can contribute to a fairer selection for higher education, concludes Marcus Strömbäck Hjärne.
Download the dissertation: A fair score? Group independent validity arguments for college admission tests