Estimation in the piece-wise constant hazard rate model-when the data are grouped

An extension of a model described in Coivert and Boardman (1976) Is Introduced lor the case when the data are available onl"r at certain time Periods. Exact and approximate maximum likelihood estimators for the parameter of the piece-wise exponential model are obtained for the general case when the interval spacings are not equal. In the special case when the Interval spacings are constant within each of the"pieces"in the model, exact maximum likelihood estimates are obtained. An example from an industrial life test illustrates the procedures in the special case when only two pieces are assumed.