|
Abstract:
|
For complex survey data, the parameter estimates in a quantile regression analysis can be obtained by minimizing a weighted objective function with weights being the original design weights. However, when the complex survey sampling design is informative, the aforementioned design-weighted estimator may be inefficient. In this paper, we proposed several weight smoothing estimators for quantile regression analysis of complex survey data collected with an informative sampling design. In developing our new estimators, we incorporated non-parametric methods for modeling the weight functions and adopted pseudo population bootstrap methods for variance estimation. We then conducted a simulation study to compare all newly proposed methods with the original design-based method in terms of bias, standard error, mean squared error, and coverage property. Results from the simulation study showed that our proposed estimators had smaller bias and mean squared error than did the design-based estimator. We further illustrated and compared all estimators by using the 1988 US National Maternal and Infant Health Survey.
|
