This paper reports on a preliminary study conducted for gaining better insight in the complexity of students' misconceptions of representativeness. Data from 156 students (112 high school graduates and 44 students with a university degree) are presented. The overall outcome indicates a lack of ability to refer problems about specific experiments to their correct context. Some results seem to contradict part of the representativeness heuristic described by Kahneman and Tversky (1972). They might also indicate that multiple-choice tests, even with two-part questions, are not able to fully capture the deep complexity of students' misunderstandings.