Sir Ronald Fisher was a mathematician and scientist who formalized what is called the analysis of variance, a statistical technique. It is one of the two big techniques used today, the other being correlation/regression analysis. Fisher wrote “The Design of Experiments”, a book that has been a mainstay of statistics and experimental design for years. When the book reached its 21st printing, my favorite statistician, John Tukey*, wrote a review of the book, saying that when a book reaches 21, it becomes an adult and deserves special consideration.
In his book, Fisher created a story of a British woman who had consumed many cups of tea. This hypothetical woman alleged that she was so experienced and so sensitive that she could tell by tasting whether the cream* was added to the cup and then the hot tea (from a tea pot, no tea bags!) or the cream was added second and the tea first. Fisher showed that the math involved supported the idea that the lady’s claim should be tested by presenting 8 partially filled cups to her to taste and sort, where 4 of them were prepared one way and 4, the other. The cups would be randomly set on a tray and her job was to sort them correctly. 100% correct sorting would be sufficiently improbable by chance alone that it would be rational to conclude that the lady did indeed have the ability she claimed.
This general plan, sorting two sorts of examples into the correct groups, is applicable to many tests and claims. Lynn once made homemade Kahlua but found it a poor substitute. My friend and I thought it was good stuff and claimed she was merely prejudiced. She knows the lady-tasting test and challenged us to take a test of the real stuff and her recipe. We both took the test and we both failed. She took the test and passed.
Students in my classes have set up the test between margarine and butter, between a cola beverage and a white soda, and many other comparisons.
* A note: Tukey’s name makes people think of the word “turkey”. He was a very brilliant man and far from a turkey. Still, it is worth knowing that there is a paper “A Quick Test According to Duckworth’s Specifications” by John Tukey. That is true and the best stat humor I can manage.
*I received a note from a British professor stating that only an ignorant Yank would write about tea and cream when any Brit knows it is tea and MILK.
