Demystifying the Machine: Small group activities to reinvent statistical methods from scratch


Cassandra Pattanayak (Wellesley College)


Abstract

As statistics education increasingly embraces both technology and active learning, what should students actually do when they are working in groups during class? Though jointly running computer code has its place, another strategy for active learning involves asking small groups of students to turn off their screens and pretend that they are the first ones to approach a common statistical problem. This presentation describes and demonstrates a series of in-class group activities designed to engage students in demystifying statistical methods before relying on software forevermore. For example, the pros and cons of hypothesis testing may be more obvious after students design their own strategies to find out whether a friend's party trick is real; the idea of an interaction in a multiple regression model may be more clear after attempting to write down an equation that describes a provided scatterplot; a short struggle to guess at a missing cell in a small spreadsheet prepares students to ask the right questions when using statistical software to impute missing values. The particular activities described come from a second-level applied statistics course, though the exercises could be adapted for an introductory course. I will also discuss strategies for assigning students to groups to maximize learning and inclusiveness.

OkCupid Dataset:
https://github.com/rudeboybert/JSE_OkCupid


Recording

register