William Gosset William Sealey Gosset

William Gosset

William Sealey Gosset was one of the leading statisticians of his time, particularly with his work on the concept of standard deviation in small samples. His theories which were published under the name of "student" are still used today in both the study of statistics and the practical application.

Gosset was born on June 13, 1876, in Canterbury, England to Colonel Frederic Gosset and Agnes Sealy Vidal. Gosset was well educated from the beginning first at Winchester, a prestigious private school, then at New College at Oxford. He received his degree in mathematics in 1897, followed two years later by a degree in chemistry (O'Connor and Robertson). It was the combination of these two fields of study that gave Gosset a career and an opportunity to create his theory.

Upon graduation, Gosset was hired as a chemist by the Arthur Guinness and Son Company in Dublin. Working in the brewery, required Gosset to constantly attempt to find the best varieties of barley for use in the production of Guinness. This was a complicated procedure of taking small samples to determine the best quality product. Gosset continuously played around with the results of various samples of barley in order to find ones of the best quality with the highest yields that were capable of adapting to changes in soil and weather conditions. Much of his work was trial and error both in the laboratory and on the farms, but he also spent time with Karl Pearson, a biometrician, in 1906-07, at his Galton Eugenics Laboratory at University College (O'Connor and Robertson).

Pearson assisted Gosset with the mathematics of the process. Gosset published his findings under the name of "student" because the brewery would not permit him to publish. The brewery feared that trade secrets would get out if information about the brewing process was published. Consequently, Gosset had to assume a pseudonym even though his information would not have impacted the business in the way the brewery was concerned (O'Connor and Robertson).

Gosset published his work in an article called "The Probable Error of a Mean" in a journal operated by Pearson called Biometrika. As a result of Gosset's pseudonym, his contribution to statistics is called the Student t-distribution. Gosset's work caught the attention of Sir Ronald Aylmer Fisher, a statistician and geneticist of the time. Fisher declared that Gosset had developed a "logical revolution" with his findings about small samples and t-distribution (O'Connor and Robertson).

In his work with the barley for the brewery, Gosset was concerned with estimating standard deviation for a small sample. A large sample's standard deviation has a normal distribution. However, Gosset did not have the luxury of working with large samples. He had to find a way to determine the standard deviation for a small sample without having a preliminary sample to make an estimate. Gosset developed the t-test to satisfy this need. Gosset's conclusion was that "small samples taken from an essentially normal population have a distribution characterized by the sample size" (Calkins). Gosset determined what the properties would be for a small sample in relation to a larger sample.

Gosset applied his discoveries to his work for the brewery which raised the question of what his focus was in his work as he was a theorist to a certain extent, but also a practical man who applied the knowledge that he developed. Gosset must have recognized the importance of accurate statistical information to the process of brewing as he had a statistical assistant for many years in his lab and ran a statistics department as part of the brewery (O'Connor and Robertson).

After Gosset had worked for many years developing the practical application of his theory, he was involved in a terrible car accident in 1934 which left him incapacitated for many months. During this time, he had the opportunity to continue to work on his statistical work. He recovered enough after a year to move to London where he became the head brewer and scientist of production at a new Guinness brewery. Gosset continued to publish the results of his statistical findings while working in London. He did not hold his position there long as he died in Beaconsfield, England, on October 16, 1937 (O'Connor and Robertson).

Gosset's legacy to the world includes more than his scientific discoveries. He will always be remembered for those, but amongst his peers and friends he was remembered as a kind and gentle man. A friend described his as "very kindly and tolerant and absolutely devoid of malice. He rarely spoke about personal matters but when he did his opinion was well worth listening to and not in the least superficial" (McMullen as quoted in O'Connor and Robertson).