An official website of the United States government
United States Department of Labor
United States Department of Labor
The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information,
make sure you're on a federal government site.
The site is secure.
The
https:// ensures that you are connecting to the official website and that any
information you provide is encrypted and transmitted securely.
Office of Survey Methods Research
SHARE ON:
More Information
Published Year: 2023
Survey/Program(s): Current Employment and Statistics (CES),
Catalog: Statistical Survey Paper
Download PDF
Testing of Weight Smoothing Models in the Current Employment Statistics Survey with SAS and R
Abstract
Sample units with extreme values can have undue influence on survey estimates. This is particularly the case when those sample units are associated with large design weights and the sample size is small. The extreme values with large design weights can disproportionately affect survey estimates and impact their stability. Using establishment survey data from the Current Employment and Statistics (CES), we explore methods for weight smoothing to reduce weight volatility and improve the stability of the survey estimates. This paper extends the previous work of Gershunskaya and Sverchkov (2014), in which they considered several models for weight smoothing, e.g., LOESS curves, penalized B-splines, and Bayesian models and compared weighted estimates from those methods to true values. We consider an additional set of methods to accomplish the same goals. These include using the CES Robust Estimator, mixed random effects, bagging, and high-performance split modeling. We compare weighted estimates from these methods to full administrative counts from the Quarterly Census of Employment and Wages (QCEW).