This article, at least for me, brought up a very interesting question: is visualizing data structures the best way to them? Please hold all the tool comments for later, but this question has particular significance for cs016.

This article was from 1993, and it stated that at that point in time there was no empirical evidence to demonstrate the value of visualization. As a result, the authors decided to create their own test to see if visualization gave students an advantage over students who only had textual descriptions.

This is where I started to get a little confused. The test was conducted on graduate level students, and the data structure they described sounded rather confusing. My question was with such a difficult concept, and given such a short amount of time to study the algorithm, could anything have really helped that much? The authors expected the animation to give students a serious edge, but with only 15 minutes to play with it I doubt that students could have got much out of it. I think more empirical evidence and better testing is needed to make a concrete recommendation.

Personally, I feel that visualization is a useful tool that helps students to actually see what is going on with the data structure. I do feel, however, that sometimes too much emphasis is placed on the picture being generated, as opposed to what is really going on underneath the picture. I think that an algorithm's course can benefit greatly from visualization, but some textual representation is also necessary.

Ah! You reminded me of a possible flaw in their hypothesis testing design. Since this experiment was performed on grad students who had gotten past at least four years of learning data structures and algorithms without a visualizer, how could the test be accurate for all people learning algorithms? It could be possible that students who would have learned better using the visualizer left CS before grad school. A better test would have been on students learning algorithms for the first time! Just my thought.

Lucas:

I'd say that the best test would have to be two different groups -- one a set of "virgin" CS students, and one a set of grad students or industry professionals that did not learn via a vizualizer the first time around.