Efficient sampling is important in many statistical applications. When sampling units is expensive or time-consuming, procedures are needed which limit the size of the sample taken while maintaining a high level of precision. Adaptive sampling is one way to efficiently sample from a population. In adaptive sampling, the sampling design is modified as data is collected. Another sampling method is Neyman allocation, which is known to give the optimal sample size allocation in stratified sampling under certain conditions. The difficulty of performing Neyman allocation in practice is that it requires some knowledge of population parameters.
We propose a sampling procedure that combines adaptive sampling with Neyman allocation. Our procedure allows one to perform Neyman allocation approximately with the benefit that no prior knowledge of population parameters is needed.
|