On representativeness, informative sampling, nonignorable nonresponse, semiparametric prediction and calibration

Informative sampling refers to a design for which the sample selection probabilities depend on values of model outcome variable. In such cases holding data is different from population data. Similarly, nonignorable nonresponse mechanism in response probability depends value missing For this paper, we study, within modelling framework, semi-parametric prediction finite total by specifying distribution units under informative and nonresponse. This most general situation surveys other combinations informativeness mechanisms can be considered as special cases. Furthermore, based relationship between distribution, introduce new measure representativeness set test sampling, jointly. Finally, calibration estimator obtained when nonignorable.