Sampling Designs - Sampling Spatial Units for Agricultural Surveys - page 144

Although they were developed in the 1940s, stratification and multistage
sampling are still attractive solutions when planning surveys. Indeed, it
appears that the large amount of auxiliary data and the development of new
data collection methods have resulted in even more design options. While the
computational burden is less important now than in the 1940s, the simplicity
of the basic concepts of these designs and their resulting HT estimator are still
valid.
Note that the features of each design are such that they are generally used
together to improve the efficiency and quality of the estimates. A ps of the

PSUs is used to exploit the different sizes of the units, while at a second stage
we could combine information from SSUs taken using different designs and
for different purposes (such as a stratified or systematic design).
It is important to observe that our choices are not only driven by minimiz-
ing the sampling error: we should also try to minimize errors from other
sources such as the nonresponse rate, management, and organizational costs.
The main gap to be filled is that these classic designs do not exploit a
fundamental characteristic of many populations: their geographical position.
They are widely used to select samples from populations that are geo-coded
or consist of spatially defined units, but do not use this particular feature.
A better design for representing a population should take into account
these characteristics, but none of the designs described in this chapter can be
straightforwardly adapted to efficiently exploit the spatial distribution of a
population. This topic will be described in the next Chap. 7 .

π

References

Al-Saleh MF, Al-Omari AI (2002) Multistage ranked set sampling. J Stat Plan Inference
102:273-286
Bai Z, Chen Z (2003) On the theory of ranked—set sampling and its ramifications. J Stat Plan
Inference 109:81-99
Baillargeon S, Rivest LP (2009) A general algorithm for univariate stratification. Int Stat Rev
77:331-344
Baillargeon S, Rivest LP (2011) The construction of stratified designs in R with the package
stratification. Surv Methodol 37:53-65
Bee M, Benedetti R, Espa G, Piersimoni F (2011). Cut-off approach to the design of longitudinal
business surveys. In: Joint statistical meeting proceedings, statistical computing section,
American Statistical Association, Alexandria, VA, pp 2488-2500
Benedetti R, Piersimoni F (2012) Multivariate boundaries of a self representing stratum of large
units in agricultural survey design. Surv Res Meth 6:125-135
Benedetti R, Espa G, Lafratta G (2008) A tree-based approach to forming strata in multipurpose
business surveys. Surv Methodol 34:195-203
Benedetti R, Bee M, Espa G (2010) A framework for cut-off sampling in business survey design. J
Off Stat 26:651-671