Adaptive Cluster Sampling in Two-stage Sampling

Adaptive cluster sampling can be a useful design for surveying rare and clustered populations. Here we present a new development in adaptive cluster sampling where we use a two-stage design and extend the complete allocation sampling method. In the proposed new design the primary sample units are selected and, depending on the value of a preset condition, the entire primary unit is surveyed, as in complete allocation sampling. In the next step, if a second condition is met, the surrounding primary sample units are selected. We review the efficiency of the proposed design for sampling the New Zealand Castle Hill buttercups and provide unbiased estimators for the population total and sampling variance.