CHI-SQUARE TESTS-OF-FIT FOR LOCATION-SCALE FAMILIES USING TYPE-II CENSORED-DATA

Chi-square tests-of-fit that allow a versatile choice of point estimators and provide maximum degrees of freedom have been available since 1980. Unfortunately, the quadratic forms involved are sometimes difficult to compute. The formulae are formidable at first glance. However, several useful quadratic forms are quite easily computed. This paper demonstrates the use and computation of such quadratic forms by performing tests-of-fit on a type-II censored data-set for the exponential family. Atl calculations are readily performed on a hand-held engineering calculator. The technique applies in general for testing fit to location-scale families of distributions.