Viktning av delproven i högskoleprovet
2006 (Swedish)Report (Other academic)
The Swedish Scholastic Assessment Test (SweSAT) presently consists of five subtests with a total of 122 items. Each correct answer gives one point, and these points are summed to a total score. After the test, the total score is transformed to a standardized score on a scale from 0.0 to 2.0. Three of the subtests measure verbal ability (vocabulary with 40 items, Swedish reading comprehension with 20 items, and English reading comprehension with 20 items) and two subtests measure quantitative ability (data sufficiency with 22 items and interpretation of diagrams, tables and maps with 20 items). Hence the verbal part of the test amounts to 66 percent of the total test. It has been suggested that the predictive power of the test might be improved if the test was composed of two separately scored parts, one measuring verbal and the other measuring logical ability. These parts could be weighted differently for application to different educational programs. In order to examine the effect of a changed weighting of the parts of SweSAT for different subgroups of examinees a simulation study was performed. Two different weighting models were used. In one weighting model the verbal part constituted 59 percent, and the logical part constituted 41 percent of the total test. In the other model the verbal part constituted 49 percent and the logical part 51 percent. These models were used on the results of the 62 700 examinees, who took part in the test, administered in spring 2000. The results regarding gender were, as expected, that the weighting models favoured males. Regarding age groups, the younger examinees (below 25 years) were favoured. Regarding different educational groups those with low education and those with higher education were disfavoured, while the big group of examinees who had finished upper secondary school and nothing else, were favoured. And finally regarding social groups the weighting of the two parts hardly changed the results at all. The conclusion was that a changed weighting of different parts of the test would not have any unduly unfavourable consequences.
Place, publisher, year, edition, pages
Umeå: Institutionen för beteendevetenskapliga mätningar, Umeå , 2006. , 14 p.
BVM / Institutionen för beteendevetenskapliga mätningar, Umeå universitet, ISSN 1652-7313 ; 19
Research subject Education
IdentifiersURN: urn:nbn:se:umu:diva-14512OAI: oai:DiVA.org:umu-14512DiVA: diva2:154183