An Implementation Strategy for Efficient Convergence of the Lavallee and Hidiroglou Stratification Algorithm

Patricia Gunning, Jane M. Horgan, Gary Keogh,

Abstract:
The iterative procedure of Lavallée and Hidiroglou (1988) for stratifying skewed populations into a certainty stratum, where all the units are examined, and a number of noncertainty strata, which are sampled, has been found to have convergence problems. In this article we present a strategy for implementing the algorithm, which improves its convergence and in many cases results in smaller sample sizes than those obtained with the traditional implementation.

Keywords:
Boundaries, geometric progression, Neyman allocation, noncertainty stratum, power allocation, skewness