Too often students leave their first statistics course with at best a fuzzy understanding of basic statistical concepts and procedures. A disconcertingly high proportion cannot adequately describe or perform a t-test, for example, when taking subsequent courses. This suggests that a different approach to teaching and learning is necessary, particularly for graduate students who will need statistical tools in their research. Rote memorization of facts does not provide the preparation requisite for graduate research. A constructivist approach in course design could provide a learning environment in which students move beyond lower level cognitive skill development. Initial implementation of this approach has produced encouragingly positive results.