What's so good about problem-based learning?

In a systematically designed and controlled experiment conducted in a naturalistic instructional setting, we examined adult students' learning of two concepts. Two intact classes taught by the same instructor were assigned to I of 2 conditions. In I class, instruction was problem based for I concept. For a second concept, lecture/discussion was the exclusive method. In the other class, matching of concept and method (problem based or lecture/discussion) was reversed. Two forms of assessment of learning occurred 6 and 12 weeks following instruction. At the initial assessment, the lecture/discussion group showed superior learning for I concept and the groups performed equivalently for the other concept. At the later assessment, however, the 2 groups showed equivalent ability to access each of the concepts, but each group showed superior explanation of the concept for which they had experienced problem-based learning. Results support the hypothesis of integration of new information with existing knowledge structures activated by the problem-based experience as the mechanism by which problem-based learning produces its benefits.