Measurement of students’ knowledge is an essential part of the educational process. Teachers on universities often use computer-based tests to testing a large number of students in a short time. The question is, what kind of information these tests provide, and if it is possible to classify students on this basis. Praxis shows that the scalar test results in the form of simple numbers cannot be plainly interpreted as the level of knowledge; moreover it is not easy to build such tests, which detect the necessary information. In the first part of the article we present the results of pedagogical experiment focused on the difference between information obtained through the computer-based test and a teacher’s interview with the same students. Possible starting point to improve information from computer-based tests in non-scalar form is a construction of an adaptive test, adapting test items to identify knowledge similar to a conversation with a teacher. As a tool for the design of the adaptive tests we use so called study maps, which are described in the second part of the article.