On the multivariate asymptotic distribution of sequential chi-square statistics (Q1078943)

In many situations in multivariate analysis one decides which of several nested structural models appears to be ''best'' for a population of interest. The sequential chi-square test (SCT) or the sequential chi- square difference test (SCDT) are popularly used in literature. In this paper, the multivariate asymptotic distribution of SCT statistics is investigated. It is shown that the SCT statistics calculated for nested models on the same data have an asymptotic intercorrelation which may be expressed in closed form and which is in many cases, quite high, and the SCDT statistics are asymptotically independent. Some Monte Carlo evidence on the applicability of the theory is provided.