In a few sampling investigations the survey leads to non-homogeneous information, owing to the particular treated subject or to the particular typology of population involved. This happens when a part of the sample units provides reliable information, while another part provides non-reliable information on a few variables. In these circumstances the sample can be regarded as formed by two sub-samples referred to two partitions of a same population that differ in some characteristics. Let's consider, for instance, the investigations connected with some typologies of pathologies, as HIV, or those carried out on the immigrants (see centre sampling), or on particular typologies of workers. The reliable data come, respectively, from information obtained by patients registered in a medical structure, by legal immigrants and, finally, by legal workers. On the contrary non-reliable data come from information obtained by the sick people who have not yet been listed, by clandestine immigrants, and by illegal workers
