In previous articles we have considered some basic concepts related to null hypothesis significance testing. This article will discuss the origins of null hypothesis significance testing.
The origins of Null Hypothesis Significance Testing
Although the popular perception is that significance testing is a modern concept, its origins can be traced back to the 18th century. Further, the modern approach of Null Hypothesis Significance Testing (NHST) is a combination of two distinct approaches: Fisher’s significance testing, and Neyman-Pearson’s hypothesis testing. We will consider these briefly in turn.
First significance test?
John Arbuthnot (physician to England’s Queen Anne) is credited with performing the first significance test. He examined birth records in London for the years 1629-1710 and observed that the number of male births exceeded female births every year. Assuming excess male births as likely as excess female births (the null hypothesis in modern terms), he calculated the probability of the observed outcome to be 0.5^82 (or 2.067952 *10^-25). He concluded that this very small probability ruled out chance and provided evidence for the existence of God.
In modern terms, he rejected the null hypothesis at p = ½^82 significance level. Of course, Arbuthnot did not use any of the terms used in modern null hypothesis significance testing to describe his observations.