The selection process to decide admissions to higher education in Sweden is based upon the Swedish Scholastic Aptitude Test [SweSAT] and final marks from upper secondary school. The SweSAT is in many respects quite similar to, for example, the U.S. SAT and bears similar dilemmas. It is, for instance, well documented that tests of this kind tend to result in gender differences in favour of males, particularly on the mathematic parts. In the SweSAT two subtests constitute this mathematic part, namely the Diagrams, Tables, and Maps [DTM] and the Data Sufficiency [DS] subtests. As pointed out by some influential schools of thought within contemporary psychometric research, measurement devices like those mentioned above never measure a single ability. Instead performance is regarded as influenced by a number of sources at different levels of a hierarchy. The research presented here is part of a larger project aiming at investigating the internal structure of the SweSAT. Building on a previous investigation of the factorial structure of the SweSAT, a series of studies is being carried out on the dimensionality of the DTM subtest. In the first three studies, the analyses build upon the spring versions of the test taken in 1991-1995 by students 19 years of age (N = 14,431- 19,597). In a fourth ongoing study, tests administered in 2000 (N= 14,220) and onwards are being investigated. In all studies a structural equation modelling technique [SEM] was used. In short, this methodology tests hypothesized models of relationships between observed manifest variables and assumed latent variables (or constructs). The models may also include relationships between the latent variables themselves. Initially, only the DTM subtest was investigated. In addition to a general Dtm-factor and an End of test effect, a "quantitative" factor (Quant), related to items that included some kind of calculation in addition to the reading off of values, was identified. A considerable part of the problematic gender difference appeared related to this factor. In the second study a hypothesized model, including the Quant factor, was tested with the whole SweSAT. The results showed relatively strong relationships between the Quant factor and the DS subtest. The overall Dtm-factor could successfully be exchanged for a more General achievement factor for the whole test. With the aim of validating the existence of the Quant factor, in addition to gaining further knowledge of its characteristics, two more studies were launched. First the study-two model (including the whole SweSAT) was related to a somewhat modified version of a previously known model for marks according to the leaving certificate from compulsory schooling. The most striking result from this modelling was that the Quant factor in the SWESAT-model and the Mathematic/Science factor in the marks-model were strongly (positively) correlated, whereas there were no or negative relationships between Quant, on the one hand, and the General School Achievement and Language factors, on the other. In the forth study the Dtm-model was first replicated on the test administered in the spring of 2000 and thereafter related to the results of the military Enlistment Battery Test for males. As this latter test has properties that are well known with respect to what factors influence performance, the ongoing analyses are assumed to give valuable information about the properties of both the Dtm-test in general and of the Quant factor in particular. The results may help to explain the troublesome part of the gender difference in the SweSAT. Moreover, a better knowledge of the test may also contribute to the discussion as to how tests of this type could be better constructed to be of use in the selection process for higher education.