This research examines an implementation of an activity-based constructivist perspective to teach the concept of a sampling distribution of a statistic. A correct conception and understanding of the sampling distribution of a statistic is crucial for students to be able to understand and correctly interpret hypothesis tests and confidence intervals. In particular, a comparison is made between this constructivist method of instruction and a traditional transmission mode of instruction in terms of student attainment of the concept of sampling distributions. In addition, qualitative research methods were employed to gain comparative data and extensive descriptive information on learning outcomes of students involved in the constructivist/reform instructional method. In terms of an overall empirical measure of student understanding of sampling distributions, the activity-based constructivist method implemented here, promoted a deeper and more complete understanding. Such a test, however, obscures an interesting phenomenon, the activity-based constructivist strategy, counter to typical constructivist claims, does not promote conceptual understanding for all students. Qualitative analysis seems to indicate a very complex interaction concerning the epistemology the student brings to the class, the connection between the students' epistemology and the epistemology inherent in constructivist instructional methods, the content of the activity, prior educational experiences, and the social/academic atmosphere of the class/institution.