P2-05: How Adapting the M&M Experiment to the Online Classroom Yielded a Useful Messy Data Set

By Jane Oppenlander (Clarkson University)


In a 2013 Journal of Statistics Education article, Todd Schwartz describes a classroom activity for teaching one-way analysis of variance using M&M’s candy. This poster describes how this activity was adapted for use in an introductory MBA statistics course delivered online in an asynchronous format and the unexpected by-product of a messy data set when online and classroom results are merged.

The instructor maintains a single data set updated to contain the experimental results from each successive class. This data set is used throughout the course to teach other topics such as data visualization and methods for proportions. Differences in how the experiment is conducted in the classroom and online settings yields a data set that contains results for different types of experimental units (e.g., different sized packages or varieties of M&Ms) and missing values. This offers the students an opportunity to actively learn data science skills, particularly, how to select appropriate records to address a research question. Experiences will be shared describing the challenges and benefits associated with both the classroom and online implementation of the M&M experiment.