Two-phase sampling is the sampling scheme including two distinct phases, in the first of which the information about the particular variables of interest is collected on all the members of the sample, and in the second, information about other variables is collected on the sub-sample of the individuals in the original sample. An instance of where this type of sampling procedure may be useful is when estimating prevalance on the basis of results given by a fallible, but inexpensive and simple to use, indicator of the true disease state of sampled individuals.
The diagnosis of the subsample of the individuals may then be validated through use of an accurate diagnostic test. This kind of sampling process is often wrongly referred to as the two-stage sampling which in fact includes a completely different design.