Post-Stratification Estimators and Related Techniques.

摘要

It is the objective of this dissertation to develop techniques in current use in sample survey theory which employ auxiliary information in the form of so-called concommitant variables. The main section of this dissertation is concerned with a situation in which the auxiliary variables are used to classify the sample units into 'post strata'. In this situation the auxiliary variables will be known only after sampling and cannot be used for the construction of strata to be sampled in prescribed quotas. The second aspect is concerned with so-called 'regression estimates' and here the customary large sample theory of linear regression estimators with one concommitant variable is extended to a second order (parabolic) regression. The third aspect is concerned with extending the 'double ratio estimators' to 'multivariate double ratio estimaotrs' and the well known 'linear regression estimators' to 'linear regression estimators with several auxiliary variables'.