Bears in Space : Activities to Introduce Basic Ideas of Design

Two elements of the reform movement in undergraduate statistics education are greater emphasis on data production (a curricular change) and greater emphasis on classroom activities (a pedagogical change). This paper presents a sequence of closely related activities, using simple equipment, for introducing students to the basic concepts of experimental design. The activities take as their nominal goal to understand the factors that influence the distance traveled by soft candy bears catapulted from a homemade launcher. They introduce students to variability and confounding, the need for a protocol, for replication and randomization, blocking as a strategy for reducing variability, and factorial crossing as a strategy for studying interaction. The analysis can be kept simple and informal, but the data also lend themselves to a more ambitious analysis, including, for example, both formal ANOVA and related diagnostics; in particular, the variation in launch distances is proportional to the mean distance, suggesting a transformation to logarithms.