The objective of the present paper is to propose a family of separate-type
estimators of population mean in stratified random sampling in presence of nonresponse
based on the family of estimators proposed by Khoshnevisan et al.
(2007). Under simple random sampling without replacement (SRSWOR) the
expressions of bias and mean square error (MSE) up to the first order of
approximation are derived. The comparative study of the family with respect to
usual estimator has been discussed. The expressions for optimum sample sizes
of the strata in respect to cost of the survey have also been derived. An
empirical study is carried out to shoe the properties of the estimators.