Editor--Part 2 of this article can be found here.

In some industries a few test batches will be produced prior to going into production. When this happens, a critical question is: “Are all of the test batches alike?” With only one value per batch, how can we compare a set of three or more values to see if one of the values is different from the others? This article will provide an answer by presenting a new test for homogeneity.

The problem here is substantially different from most of the problems addressed by statistical tests. Here we will want to use the test batches to characterize future production, but before we can meaningfully interpret even the simplest statistics in this way, we will need homogeneity between the test batches.

 …