Multivariate Boundaries of a Self Representing Stratum of Large Units in Agricultural Survey Design

Optimal Stratification and Allocation for the June Agricultural Survey

“…Self-representing strata are also found in Benedetti et al (2010) and Benedetti and Piersimoni (2012), however the approach presented can exceed three strata and is not restricted to highly asymmetric populations. Instead, it relies on the observation that, given an initial allocation and stratification, then a sequence of exchanges can be taken along primarily stratum boundaries to attain a more optimal design.…”

“…Optimization of this problem is performed by a stochastic optimization method known as simulated annealing (Metropolis et al 1953). This optimization method is also used in Benedetti et al (2010) and Benedetti and Piersimoni (2012). This stochastic optimization method is a metaheuristic that uses a Monte Carlo method to obtain an optimal solution by generating a sequence of possible solutions that slowly converge to an optimal solution.…”

“…This method is commonly known as the cum ffiffi ffi f p method (Sa ¨rndal et al 1991, Section 3.7) (Horgan 2006). Lavalle ´e and Hidiroglou (1988) and the multivariate extensions in Benedetti et al (2010) and Benedetti and Piersimoni (2012) provide optimal designs under CV constraints, but restrict the strata to either two or three stratum. These stratum in Benedetti and Piersimoni (2012) include a census (take-all) and sampled (take-some) strata and do not restrict the sample size.…”

“…Lavalle ´e and Hidiroglou (1988) and the multivariate extensions in Benedetti et al (2010) and Benedetti and Piersimoni (2012) provide optimal designs under CV constraints, but restrict the strata to either two or three stratum. These stratum in Benedetti and Piersimoni (2012) include a census (take-all) and sampled (take-some) strata and do not restrict the sample size. Benedetti et al (2010) included a third (take-none) stratum for the purposes of cut-off sampling.…”

“…These approaches are designed for highly skewed populations, exploiting the similarity of the underlying population to a geometric progression (Gunning et al 2004). Benedetti and Piersimoni (2012) introduced a method for stratification which uses multiple administrative variables. This method, which is motivated by the Lavalle ´e and Hidiroglou method, partitions the population into two strata, one which is sampled and one, which is a take-all stratum.…”