Asymptotical bias of total predictors in poststratified sampling

Poststratification is a current technique in survey sampling. It is used when a regression superpopulation model generates the observations. This technique is analyzed through two predictors which are suggested by the estimators studied previously by Pfefferman-Krieger. A third predictor is proposed. Their asymptotical conditional design biases are obtained. They are used, as is currently done in the literature, for evaluating their behaviour.