Omnibus goodness-of-fit tests for count distributions

A consistent omnibus goodness-of-fit test for count distributions is proposed. The test is of wide applicability since any count distribution indexed by a $k$-variate parameter with finite moment of order $2k$ can be considered under the null hypothesis. The test statistic is based on the probability generating function and, in addition to have a rather simple form, it is asymptotically normally distributed, allowing a straightforward implementation of the test. The finite-sample properties of the test are investigated by means of an extensive simulation study, where the empirical power is evaluated against some common alternative distributions and against contiguous alternatives. The test shows an empirical significance level always very close to the nominal one already for moderate sample sizes and the empirical power is rather satisfactory, also compared to that of the chi-squared test.