10th International
Conference on
Teaching Statistics
8 – 13 July 2018
Kyoto, Japan
This paper is from Session 1C: Statistics Education: What, how and with whom?
Full topic list
which comes under Topic 1: Statistics education: Looking back, looking forward

Paper 1C2 (Friday 13th, 14:00-15:30)

Confounding and Cornfield: Back to the Future



Most students see less value after taking intro statistics than before. Most students are in disciplines that deal with observational studies and confounding. Introductory statistics doesn't deal with observational studies. Most intro statistics textbooks never mention confounding. Confounding wasn't one of McKenzie's top 30 concepts. Why are we so silent? Perhaps we aren't comfortable teaching confounding. To change the future, we need to go back to when Jerome Cornfield argued that smoking caused cancer. Fisher disagreed arguing it might be caused by a genetic confounder. Cornfield refuted Fisher by deriving the Cornfield conditions: one of the greatest contributions of statistics to human knowledge. We need to teach confounding and the Cornfield conditions so students will appreciate statistics.

Download/view the paper