The Lady Tasting Tea

April 2, 2001
The development of statistical modeling in primary research is the underreported paradigm shift in the foundation of science. The lady of the title's claim—that she could detect a difference between milk-into-tea vs. tea-into-milk infusions—sets up the social history of a theory that has changed the culture of science as thoroughly as relativity did (the lady's palate is analogous to quantum physics' famous cat-subject), making possible the construction of meaningful scientific experiments. Statistical modeling is the child of applied mathematics and the 19th-century scientific revolution. So Salsburg begins his history at the beginning (with field agronomists in the U.K. in the 1920s trying to test the usefulness of early artificial fertilizer) and creates an important, near-complete chapter in the social history of science. His modest style sometimes labors to keep the lid on the Wonderland of statistical reality, especially under the "This Book Contains No Equations!" marketing rule for trade science books. He does his best to make a lively story of mostly British scientists' lives and work under this stricture, right through chaos theory. The products of their advancements include more reliable pharmaceuticals, better beer, econometrics, quality control manufacturing, diagnostic tests and social policy. It is unfortunate that this introduction to new statistical descriptions of reality tries so hard to appease mathophobia. Someone should do hypothesis testing of the relationship between equations in texts and sales in popular science markets—it would make a fine example of the use of statistics. Illus.