Sex differences on the Swedish Scholastic Assessment Test

The question of cognitive sex differences has been brought to the fore by the urge for equality in various domains, including higher education. The stark discrepancies in sex ratio across STEM, medicine, law, and the... [ view full abstract ]

The question of cognitive sex differences has been brought to the fore by the urge for equality in various domains, including higher education. The stark discrepancies in sex ratio across STEM, medicine, law, and the humanities may be more related to differences in interests, cognitive profiles, and distributional widths of cognitive ability, rather than psychometric g or full-scale IQ. Here we address this question using data from the Swedish Scholastic Aptitude Test (SSAT), which is annually taken by about 100 thousand individuals. Initially intended in 1977 as a second chance to qualify for tertiary education for those who had not attended or failed college, it became open to anyone in 1991. Having changed and developed over the years, it now consists of 8 subtests covering 4 verbal tests (3 in Swedish and 1 about English comprehension), and 4 quantitative tests including logical and mathematical problems. Males tend to perform 0.3-0.5 SD better than females overall on the SSAT, consistent with the general pattern that males tend to (a) get poorer grades, (b) have about the same level of knowledge, and (c) perform better on tests generally. Such a difference, although relatively large, is difficult to interpret - especially given that the test is voluntary and therefore could be biased by self-selection. It can also be taken as many times as one likes (it is given twice per year). The present analysis therefore attempts to control for such factors, using a particular sample for which there is data on parents’ background (N ~ 15,000), so as partly to control for heritable intelligence. Remaining differences in both profile and distribution, as well as a tendency for males to perform better on more difficult items, would be inconsistent with exclusive sociocultural causation.