ABSTRACT
This chapter considers a variety of techniques which are appropriate as tests of fit when only a certain portion of the random sample from a continuous underlying distribution is available. The method of the previous section can be applied directly in any situation where the data consist of some known subset of order statistics from a random sample. The method of probability plotting extends easily to multiply right-censored samples; however, the computation of quantile probabilities is more complicated. Probability plots for the individual components were constructed using several common life distributions. There are other more complicated types of multiple censoring which can arise in practice. B. W. Turnbull and L. Weiss present an omnibus test for a composite null hypothesis based on the generalized likelihood ratio statistic. H. J. Chen presents a correlation statistic as an omnibus test for the composite hypothesis of exponentiality in the presence of random censoring.
