Nonparametric Survey Regression Estimation in Two-Stage Spatial Sampling

ABSTRACT
A nonparametric model-assisted survey estimator based on local polynomial regression is extended to incorporate spatial auxiliary information and two-stage sampling designs. Under mild  assumptions, this estimator is asymptotically design-unbiased and consistent. Simulation studies  show that the nonparametric regression estimator is competitive with standard parametric  techniques when the parametric specification is correct, an outperforms those techniques when the  parametric specification is incorrect. The methodology is applied to water chemistry data from the EMAP Northeastern Lakes Survey.
 

College of Natural Sciences