I agree with this article that attempting to recognize common mistakes by novice programmers is an excellent (and difficult) undertaking. However, I think it was very good of this article to attempt to approach the question with research and statistical evidence, rather than just anecdotal evidence of how the author/professor/TA happened to learn, and what problems they had.

One thing which struck me is that their research was, as they acknowledged, based on a set of students who learned programming in a traditional procedural manner. They learned declaration/assignment, then conditionals, then loops, etc. I am curious as to how different the common bugs and problems are among intro students who are taught in a different manner. For example, does CS15 teaching objects first gear the students to different bugs and problems than the procedural approach? How about starting students with a functional programming language, such as LISP? I think there would be some commonality among all problems, and the difference between construct-based problems and plan composition problems would probably hold in all cases. However, it would be interesting to see a study comparing the bugs of students taught with different styles.

Another thing that struck me was the number of typos in this document. If this is funded by government dollars, they could have at least put it through a spell checker!

My Comments: One comment about the testing: what did it really test? It struck me that the testing was geared to see how much syntax students could remember. Trying to right a thoughtful program I realized I don't remember GP at all, but does that mean that I make mistakes because I'm a novice, or because I can only remember my name because its written on my shoe?