As. Hedayat et Wz. Robieson, EXCLUSION OF AN UNDESIRABLE SAMPLE FROM THE SUPPORT OF A SIMPLE RANDOM SAMPLE, The American statistician, 52(1), 1998, pp. 41-43
It is shown that for any finite population of size N and any sample si
ze n greater than or equal to 3, we can have a survey plan of size n t
hat assigns a probability of zero to an undesirable sample of size n a
nd yet the plan is equivalent to a simple random sample plan of size n
without replacement.