Improving students' problem-solving skills is a major goal for most science educators. While a large body of research on problem solving exists, assessment of meaningful problem solving is very difficult, particularly for courses with large numbers of students in which one-on-one interactions are not feasible. We have used a suite of software tools and pedagogies of small-group student collaborations to assess both student problem-solving strategies and student abilities as they change over time. This paper reports the use of these tools to probe the effectiveness of using small group interaction to improve problem solving. (Contains 1 table and 5 figures.)