GMO Statistics Part 21. The tests Gilles-Eric rejects part 2.

In the previous post we introduced the reader to the statistics of the discrete counted event, such as the numbers of deaths occurring in a particular time interval. We gave the example of the Poisson probability distribution, which only relates to whole number variables such as 1, 3 or 42. Here we provide one example, from a standard medical statistics textbook, showing how it is customary in medical science to provide confidence intervals on such estimates of rates of disease, so that their values may be carefully interpreted and compared with one another while properly recognising the extent to which chance variation may influence the actual numbers counted.

(From Essential Medical Statistics, B R Kirkwood and J A C Sterne. 2003, 2nd Ed. Blackwell.)

The value d in the above image refers to the number of occurrences of the counted event.

Elsewhere, this standard medical text explains that the standard error of the rate so estimated is proportional to the square rout of the number of occurrences given a uniform counting interval for making comparisons. This means that a small number of occurrences, such as 3 deaths in a group of 10 rats, has a wide uncertainty.

This is indeed, essential medical statistics that Gilles-Eric Seralini rejects.