If you’ve ever had to measure the effectiveness of a new product or process design, you know how valuable significance testing can be. But did you know how beer made that testing much easier?

Relevance and significance are two of the most important concepts in modern statistical analysis. Relevance helps us to understand the connection between the research and the problem to be solved and significance tells us if that research represents a change that is greater than can be attributed to chance. While relevance is typically determined through methods of logic, significance is a purely mathematical exercise.

In most statistical trials, managers ask for large numbers of data points in order to reach a desired confidence level. Traditionally the magic number has been 30 data points. (Actually, there is no magic to the number 30. It just seems that in most distributions the standard error reduces to about 1.0 somewhere between 20 and 30 data points.)

 But what about when such large sample sizes are impractical?

Anyone who has ever taken an introductory class in statistics has learned to use the Student’s T-test of statistical significance. The “Student’s T-test” is one of the most basic tests of small-sample significance in statistics. But many people might be surprised to learn that it really has nothing to do with students, and almost everything to do with beer.

In 1906, William S. Gosset was a bright young chemist working for the Guiness Brewery in Ireland. Gosset developed a small-sample method for measuring the deviation of means in the production of Guiness’ dark beers. This allowed for accurate tests of significance without the need for excessively large sample sizes.

His method was so successful he submitted it for publication in Biometrika, a professional journal published by his friend and professor Karl Pearson. And here, believe it or not, is one of the few times in history where a story about math gets interesting.

A popular version holds that Guiness had a policy against its employees publishing their work. So Gossett compromised with his employer and used the pen name he had published under before: Student. Some still like to contend that Gosset never let his employer in on the fact that he was publishing and submitted his work under cloak and dagger. Probably not true, but makes for intriguing reading.

It is far more likely that Gosset published under Student because Guiness did not want its competitors to learn that it was using statistical analysis as a part of quality control. This was a new concept at the time and the brewery probably wanted to keep it under wraps as long as possible. That did little to diminish the drama, however, as the new formula only added to the animosity between Professor Pearson and one of his chief rivals R.A. Fisher, with whom he continued to bicker for many years.

And it all started over a glass of beer.

So, the next time you are tasked with testing the significance of a difference in means with a small-sample data trial, it might be appropriate to raise a glass of beer to good Mr. Gosset.