Abstract
We investigate three alternative models for estimating the mean of a population using double sampling survey techniques. One estimator was found in the range science literature (Cook and Stubbendieck, 1986), another is the estimator presented by Cochran (1977). The third estimator uses method-of-moments estimators with measurement error regression models. Simulation studies suggest that the measurement error model does not work well when the slope is appreciably different from unity. Delta method variance estimators of the measurement error model may give negative variance estimates under these circumstances. The other estimators have better small sample performance (both are approximately unbiased, and have similar variances), but the two estimators have very different estimated variances under some circumstances.
Keywords
double sampling survey
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Clason, Dennis L. and Southward, G. Morris
(1992).
"A COMPARISON OF DOUBLE SAMPLING REGRESSION ESTIMATORS,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1411
A COMPARISON OF DOUBLE SAMPLING REGRESSION ESTIMATORS
We investigate three alternative models for estimating the mean of a population using double sampling survey techniques. One estimator was found in the range science literature (Cook and Stubbendieck, 1986), another is the estimator presented by Cochran (1977). The third estimator uses method-of-moments estimators with measurement error regression models. Simulation studies suggest that the measurement error model does not work well when the slope is appreciably different from unity. Delta method variance estimators of the measurement error model may give negative variance estimates under these circumstances. The other estimators have better small sample performance (both are approximately unbiased, and have similar variances), but the two estimators have very different estimated variances under some circumstances.
