The pdf document Statistical notes on rifle group patterns gives a general discussion of the statistics of rifle group patterns, models for their evaluation and testing including a discussion of the statistical power of the tests, and considerations of powder throw variations. It is not a tutorial, but rather a collection of statistical notes and tables that are not available elsewhere to my knowledge.

A shot pattern is well described by two independent normal distributions representing the vertical and horizontal dispersion of shots. The variances of the two distributions are not always equal, since gravity acts only vertically. In theory, the dimensions may be correlated, but in practice, the data seems to be well described by assuming independence.

The accuracy of the shot pattern is not usually of much interest, since that may be adjusted by changing the aim point either by scope adjustment or by a visual offset.

The precision, is however of great interest, and several statistics appear in the literature. Six statistics are described in a manual by Frank E. Grubbs They are defined and tabled here. The radial standard deviation, RSD, and the mean radius, MR, require the precise location of individual shots, which presents a problem on badly torn targets, making them impractical. The other four statistics, the radius of the covering circle, RC, the figure of merit, FOM, which is an average of the ranges in the two directions, the diagonal of the covering box, D, and the extreme spread, ES, are more practical. The efficiencies of the various statistics differ, but the differences are rather small, so that the less efficient and more practical statistics accrue no great loss in their use.