Geoscience Reference
In-Depth Information
adaptive allocation
to distinguish two categories. Adaptive searching refers
to designs such as adaptive cluster sampling, by which the neighborhood is
searched. In contrast, in adaptive allocation, extra effort is initiated once a
collection of units has been sampled (e.g., the stratum or the primary unit).
The distinction between the two classes is based on where and when the
decision to allocate extra effort can be made: immediately after an individual
sample unit is measured or once a collection of units has been completely
sampled. Similarly, the distinction between two-stage and two-phase sam-
pling can be confusing. In this chapter, in two-stage sampling a selection of
primary units is sampled, and in the second stage, a sample is taken within
each. In two-phase sampling as discussed in this chapter, an initial sample
is taken, and on the basis of information from that sample, additional effort
is allocated in the second phase. In adaptive searching, the decision to con-
duct the second phase of sampling occurs concurrently with the first phase.
In adaptive allocation, the decision to conduct the second-phase sampling
occurs after the first phase.
All the designs discussed can be efficient, giving estimates of popula-
tions that have lower variance than the conventional design without adap-
tive selection. However, as with conventional sampling, the survey must
be designed carefully to realize these gains in efficiency. Practical consid-
erations will limit how complex the field protocol can be, and as any field
biologist knows, there are advantages in keeping things simple. The final
design chosen for a particular survey will be a balance between theoretical
efficiencies and practical realities.
