Learning styles are more and more incorporated in technology enhanced learning and a lot of research work is done in this area. For example, systems are developed which provide adaptivity according to the learning styles of students; relationships to students' performance and other characteristics of students such as cognitive traits are investigated, and techniques are designed to derive the learning styles from the behaviour of students during an online course. The more information about learning styles is available and the more detailed the description of learning styles is, the better such approaches can work and can be investigated. The aim of this paper is to analyse data about learning styles with respect to the Felder-Silverman learning style model (FSLSM), in order to provide a more detailed description of the learning style dimensions. Therefore, we used linear discriminant analysis in order to detect the most representative characteristics of learning styles as represented in the gathered data. Furthermore, we analysed how representative these characteristics are for the specific learning style dimensions. For cross-validation, we conducted empirical frequencies analysis as well as correlation analysis. As a result, we provide a more detailed description of the learning style dimensions of FSLSM. This description is especially important when learning styles are incorporated in technology enhanced learning.