If you make anything to sell on a big scale, you need to test what you make. Not testing leaves you exposed to losing customers and lawsuits. Testing is expensive and it can be too time-consuming to test every item you make. There are clear limits. You can’t crash-test every car you make. You can’t even test every jar of pickles you sell. You can radically reduce the likelihood of putting out a bad product by testing one item in each lot of a convenient size.
How do you know just how representative of each whole lot is your set of single samples? Statistics comes to the rescue, for many, many businesses as well as many, many lab experiments, sampling of concrete or metal, and onward. The study of chance events started with well-educated gamblers, notably the philosophically inclined chevalier de Mere. De Mere posed a problem picked up by famous mathematicians Pierre de Fermat and Blaise Pascal in 1654.
A goodly number of people developed statistical testing over the years. One person of note created the most-used test ever, the t-test. William Gosset was the experimental brewer at the famed Guinness Brewery. If you experiment with beer you need to know the odds your batch is good pretty fast and simply. Now, hundreds of thousands of scientists around the world know and use Student’s t-test, under the pen name he used.
This has been an outreach activity of the Las Cruces Academy, viewable at GreatSchools.org.
Source: Scientific American, July/August 2024, pp. 88-89.
