Empirical Studies of the Value of Algorithm Animation in Algorithm Understanding

摘要

A series of studies is presented using algorithm animation to teach computeralgorithms. These studies are organized into three components: eliciting students' preferences, developing algorithm animation guidelines, and evaluating the effects of using algorithm animation in the classroom. Many systems for creating computer animations have been designed. These systems reflect the designers' confidence that visual representation is a valuable technique for conveying conceptual knowledge. Little formal experimentation has been carried out to determine whether such animations are beneficial in teaching the algorithms presented. Also, formal guidelines have not been developed for either design or use of these animations. This work addresses both concerns. Student preference for animation does not predict student performance on a post-test to measure learning. Several types of labeling (data, steps, actions) were examined and a variety of effects were found. One of these effects was that a text description of the algorithm increased accuracy on conceptual questions. Also, students are more accurate on a post-test if they are allowed to design their own data sets than if they use experimenter-defined data sets. Finally, use of the animations was evaluated in a classroom-type setting. All students were given a lecture on an algorithm. Adding a laboratory in which the subject controlled the animation and data sets led to more accurate performance than a lecture only or than a lecture with experimenter-preprepared data sets. This work has implications for the design and use of animated algorithms in teaching computer algorithms and for the design of laboratory experiences for beginning computer science courses.