Abstract:

Pitman efficiencies are used to describe the problem of choice of the number of classes in $L_{p}$-goodness-of-fit tests ($p\ge 1$) based on histogram density estimates. We consider the case where the number of classes increases with the sample size.