Simpson's Paradox is a great lesson to teach in a "statistical thinking" course (hopefully ST is taught in every intro stat course). It focuses on the problem of aggregation. There are famous examples like the Berkeley admissions data.
Other examples to track down:
1. Death rates in Maine vs. South Carolina (1970's??)--apparently there is one age group that throws everything off. In a Mosteller book?
No comments:
Post a Comment