Yesterday, I went to a panel about mathematicians teaching statistics. My department is a math/stats department so I have had the opportunity to TA many statistics classes, and I’ve really enjoyed it. The panelists all followed a similar trajectory of being forces to teach a statistics class, doing so as if it were a math class (heavily focused on definitions, equations, and procedures) and then over ten to twenty years reconstructing their class into something more focused on concepts, real world examples, and technology.

A lot of my students know how to follow the procedures of a hypothesis test quite well, but I can tell they don’t know really understand what a p-value is, and I wish I knew how to impart that understanding in the brief once-a-week discussion sections I have with them. The logic of hypothesis testing is more important, and more likely to stick with them than the details of each of the different models used in various hypothesis testing. If they do go on to use statistics in their work, they will likely be using technology, and it is the deep understanding of what a hypothesis test is that will ensure they use that technology appropriately. What I don’t think I fully appreciated before this panel was the extent to which a focus on procedures and equations can get in the way of learning statistical thinking.

The panelists have gathered a lot of useful information on this page, including links to real world data, curriculum recommendations from the MAA and ASA, and statistics teaching communities. I’ll be looking back to this the next time I get to TA (or teach!) an intro stats class.