Brian and I have watched a handful of Mythbusters lately that have used some basic statistics to either set up the test of the myth or to test the myth. I’m wondering if anyone’s ever used some of the clips in an undergraduate methods class. It seems there are lots of episodes that could but used to illustrate different types of experimental designs. But, there are also some that use statistics based on the data they collect. For instance, in this episode (see Vector Vengeance), they calculate the standard deviation across multiple cannons to determine which is the most consistent way to launch a ball. In this other episode, they are testing whether men or women, redheads or non-redheads, and cursing or not cursing explain differences in people’s ability to handle pain. They calculate means across the different groups, but stop short of a difference of means test. It seems you could (re)create the individual data to match up with their means and then do small sample difference of means tests in class.
For now, I’m making a list of episodes and saving them in my teaching undergrad stats folder in the event that I am called upon to teach that course. I also have a running list of episodes from The Big Bang Theory, including the opening scene of “The Friendship Algorithm,” when Kripke comes by and makes fun of Leonard’s experiment with 20K observations and not a single statistically significant result. Bazinga!