Abstract. The use of algorithm visualizations in computer science education is not a new thing. Although there is a firm belief that graphical representations of algorithms are learning aids, empirical studies show that what is important is what the students do with the animations rather than what they see in them. In this paper we compare to kinds of interaction: viewing animations vs constructing animations. We have conducted a controlled experiment where a group of students (n=15) had to study an algorithm and complete a knowledge test about it and a subjective opinion questionnaire. Students were randomly divided in constructing and viewing groups. Results have been measured by means of learning outcomes, efficiency issues and student's subjective opinion. Results significantly evidence that builders obtained better results than viewers.
