This thesis is concerned with the problem of analyzing field data for products with multiple modes of failure. Field data are those in which the failure time and the cause of failure are observed only for those products that fail in some pre-specified follow-up or warranty period. It is noted that for satisfactory inference about parameters involved in the model it is necessary to supplement the failure-record data by taking a supplementary sample of products that survive warranty time. It is assumed that lifetime for each failure cause follows exponential or Weibull distribution and is independent of the others.
Two types of field data are considered : i) There are no covariates for products and the actual operating time is different from its calendar time; ii) there are covariates for products and the actual operating time is the same as its calendar time. For these two cases, pseudo maximum likelihood estimators for the parameters of lifetime distribution are obtained and their asymptotic properties are studied. Nemerical examples are given and small sample properties of the estimators are also investigated.