SAYING THE SAME (OR A DIFFERENT) THING: HOW SHAPE AFFECTS IDEAS ABOUT DISTRIBUTION IN A SOFTWARE EXPLORATION ENVIRONMENT

Educational software for statistics and data analysis provides a variety of tools for seeing and expressing ideas about data distributions. However, the ideas that learners find important to express often depend on an interaction between software and the shape of the distributions themselves. In this interview study of teachers participating in the VISOR professional development program, we investigate how distributional shape (symmetric or skewed) and choice of software tool (TinkerPlot or Fathom) affect how teachers discuss data distributions when comparing groups. We find teachers' confidence is increased when different measures or ways of viewing data "say the same thing," which more often holds true with symmetric distributions. When these seem to conflict, typically with skew distributions, teachers work to understand the measures themselves, and introduce new ways of characterizing data, so that they can make coherent sense of the distributions. The paper introduces a distinction between rule-driven and value-driven measures which we find important in understanding teachers' analytic methods.