COMPARISON AMONG SEVERAL DISTRIBUTIONS

The researcher may need in some conditions a comparison among more than two distributions. The goodness-of-fit tests constitute a useful tool also in this case. Of course when we start with different kinds of distributions we have to convert some of them into the other kind. If there is not a kind preferred, a general rule is to convert all the distributions into the most numerous kind.

Once one can deal only with distributions of the same kind, the process goes on with an iteration of single goodness-of-fit tests performed on couples of distributions. With this technique we obtain some groups of distributions: those that are similar and those that are dissimilar. It is always possible to match those distributions that turn out to be similar.


Barbara Mascialino - Last update: 03/03/2006