|
Activity Number:
|
513
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #309452 |
|
Title:
|
Optimization of Nonproportional Samples
|
|
Author(s):
|
Raimund K. Wildner*+
|
|
Companies:
|
GfK Group
|
|
Address:
|
Nordwestring 101, Nuernberg, D 90319, Germany
|
|
Keywords:
|
Stratification ; Weighting ; Sampling ; Optimization ; Effective sample size
|
|
Abstract:
|
The well-known optimum distribution of stratified samples by Neyman-Tschuprow minimizes the sampling variance of the total mean of one variable. In practical surveys however, there are many variables that are to be reported with reasonable small variance not only for the total but for numerous segments. A solution is presented that minimizes the variance of the weighted sum of multiple means. Examples are shown that demonstrate how the solution behaves under different conditions. Furthermore, disproportional surveys have to be weighted to get unbiased results. An important parameter to evaluate the quality of a weighted sample is the effective sample size. But the well known formula for the effective sample size is valid only for proportional samples. The optimization formula can as well be used to derive a formula for the effective sample size for non proportional samples.
|
