ABSTRACT
In survey sampling two distinct types of studies are recognized. One is descriptive and the other is analytic. In the former on a finite population for the respective units of it, one or more real variables are supposed to take values which are totally or partially unknown or may be fully known as well to start with. Errors in estimation due to sampling are recognized and attempted to be assessed to the extent possible. In the context of finite populations such observations are supposed to be sample observations produced by simple random sampling with replacement. But since from finite survey populations samples may be chosen in more general and hence complex ways the analytic results based on assumed simple random sampling with replacement selections get distorted when actually the observations are ascertained by dint of complex sample selection procedures.
