Allocation of sample using values of auxiliary characteristic

The optimum allocation given by Neyman (1934) is not normally feasible in practice since the values of the standard deviations of the characteristic under study are not known. In this paper, it is shown how the values of auxiliary characteristic linearly related to the study variable can be used in the allocation of the sample. It is also found that proportional allocation can be more efficient than an approximation to Neyman's allocation by using estimates of standard deviations of the study variable from a previous survey or approximations to them from some variable related to the study variable.