Optimal solutions in controlled selection problems with two-way stratification
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Sun Woong Kim, Steven G. Heeringa and Peter W. SolenbergerNote 1
When considering sample stratification by several variables, we often face the case where the expected number of sample units to be selected in each stratum is very small and the total number of units to be selected is smaller than the total number of strata. These stratified sample designs are specifically represented by the tabular arrays with real numbers, called controlled selection problems, and are beyond the reach of conventional methods of allocation. Many algorithms for solving these problems have been studied over about 60 years beginning with Goodman and Kish (1950). Those developed more recently are especially computer intensive and always find the solutions. However, there still remains the unanswered question: In what sense are the solutions to a controlled selection problem obtained from those algorithms optimal? We introduce the general concept of optimal solutions, and propose a new controlled selection algorithm based on typical distance functions to achieve solutions. This algorithm can be easily performed by a new SAS-based software. This study focuses on two-way stratification designs. The controlled selection solutions from the new algorithm are compared with those from existing algorithms using several examples. The new algorithm successfully obtains robust solutions to two-way controlled selection problems that meet the optimality criteria.
Cell expectation; Probability sampling; Distance function; Optimum array; Linear programming problem; Simplex method.
Table of content
- 1. Introduction
- 2. Controlled selection problems
- 3. Desirable constraints
- 4. Optimal solutions
- 5. Non-optimal properties of existing methods
- 6. Suggested method
- 7. Software
- 8. Comparisons of algorithms
- 9. Concluding remarks