The Use of Spacings in Goodness-of-Fit Testing

One of the important steps before performing any parametric statistical analysis is to conduct a formal goodness-of-fit (GOF) testing. In this talk, we introduce somehow less known GOF testing procedure based on spacings which has been shown to have larger power in some cases in compare to the conventional Empirical Distribution Function (EDF)-based GOF methods. Indeed, GOF testing based on spacings cover a large class of tests including the Greenwood’s statistic, the Moran’s statistic and the Quesenberry and Miller’s statistic. GOF tests based on higher order spacings and some suggestions for new testing methodologies for the case of censored data will also be discussed.