Combined estimators for complex sampling

Web-surveys are more and more widespread due to their easiness and rapidity; nevertheless, the ones for which mailing lists are not used, cannot be scientifically valued as they show serious coverage errors. In fact, how it is underlined in par.2, the web-population P(w) represents only a subset of the overall population that is, in general, the one to be investigated. The problems we have to face now with regard to websurveys are similar to those that aroused at the end of the thirties with regard to telephone surveys and, as it happened then, the solution will probably be found whenthe diffusion of the web will reach the top.