ABSTRACT
The Model Assisted approach is here considered in estimating suitably the total or mean of a variable related to a “specific section” of a “survey population” on surveying a probability sample drawn from the latter. A classical design-based estimator does not suit this situation because the sample-size coming as a chunk from this specific domain may be inadequate, yielding too high a measure of error and implying a poor efficiency level. The current research deals with this situation only through model-based and more specifically the Bayesian approach.
