How a Cup of Tea Laid the Foundations for Modern Statistical Analysis
1 min read
Summary
In his 1935 book, The Design of Experiments, British statistician Ronald Fisher detailed an experiment conducted at the Rothamsted agricultural research station in which they tested whether somebody could tell the difference between tea poured with milk first or second.
The result showed that the null hypothesis—that they couldn’t—was wrong, and Fisher used this to illustrate his statistical testing theory, where a provisional conclusion is drawn from data.
However, he insisted on using a 5% cutoff for a “significant” result, and claimed that he would “ignore entirely all results which fail to reach this level.”
Though his approach, which focused on null hypothesis testing, became popular, other statisticians such as Jerzy Neyman and Egon Pearson, disagreed, preferring a decision-based approach.
The resulting feud led to decades of ambiguity, as the two competing theories were mixed up in textbooks, and the idea of null hypothesis testing and a 5% cutoff for “significant” results became ingrained in mainstream scientific research.
