Censored Observations in Randomized Block Experiments

SUMMARY Maximum likelihood methods are developed for the analysis of data from randomized block experiments, when some of the observations are censored. A modification of the maximum likelihood process, intended to produce approximately unbiased estimates of variance, is described, and the results of some numerical investigations into its efficacy are reported. Analogues of the t-test, using two alternative formulae for standard errors of differences, are considered, with some results of numerical investigations into the distribution of the t-values.