0410797 20240103182239.920060210235959.9 11 s. cav_un_epca*02576160167-7152561 (2002)57-67Elsevier goodness-of-fit statistics disparity statistics goodnes-of-fit tests cav_un_auth*0101218 Vajda Igor UTIA-B Ústav teorie informace a automatizace AV ČR, v. v. i. cav_un_auth*0212895 Gyorfi L. HU 12B IAA1075101 GA AV ČR cav_un_auth*0012783 CEZ:AV0Z1075907 We consider a class of disparity measures between discrete distributions consisting of k depending, as well as the probability themselves, on n=1,2,... and their estimates based on k multinomially distributed statistics. Using the classical results of Pearson and Tumayan, we formulate new conditions on sequences of distributions under which standardized disparities tend to the chi-square or normal laws when n increases to infinity. BB SI http://hdl.handle.net/11104/0130884 UTIA-B 20020011 2002 cav_un_epca*0257616 Statistics & Probability Letters 0167-7152 1879-2103 Roč. 56 č. 1 2002 57 67 Elsevier