Today’s class opener comes from my Advanced Placement Statistics class, but provides an important lesson for stats students of all ages. A timeplot featuring two interesting data sets, and their changes over time is featured as students enter:
That’s quite a high r value we have for two variables, autism diagnoses and organic food sales, which would not seem so closely related. In conversation with the class we discussed the importance of clear communication, and how this article could easily be summarized and misinterpreted by our local newspaper:
ORGANIC FOOD CAUSES AUTISM, RESEARCH SHOWS
Uh oh….we have a problem. And not an uncommon problem, as scientific studies which find correlations between variables are often misinterpreted as cause-effect studies. The fun site Spurious Correlations by Tyler Vigen provides some wild examples of variables with strong (sometimes eerliy strong) correlations to help frame discussions. Some fun examples –
- Divorce rate in Maine correlates with Per capita consumption of margarine (US)
- Worldwide non-commercial space launches correlates with Sociology doctorates awarded (US)
- Per capita consumption of chicken (US) correlates with Total US crude oil imports
Later, my students will be asked to read and respond to a “newspaper article” about a California school which analyzed their student data and found that student achievement correlates strongly to student height. The school’s reaction to this correlation seems dubious at best, and with good reason….it’s a fictitious article I wrote symobolize the danger of seeking cause/effect from casual relationships.
I rather like : Diagnosis of autism in child causes parents to become health food addicts ($100,000 per year per family)
Pingback: Chainsaw Review of How to Lie with Statistics, Part II | Chainsaw Communications
Pingback: Book Review: How to Lie with Statistics, Part II | Chainsaw Communications