Abstract:
|
The Current Population Survey (CPS) has measured U.S. labor force characteristics since 1940. Its variance methodology reflects a complex sample design, statistical innovations, and computational improvements. This methodology also combines best theoretical methods with techniques specific to the particular application. In this talk, we give a detailed overview of the components of CPS variance (reflecting different stages of the sample) and other related characteristics (such as design effects and generalized variance functions). We discuss the details of CPS variance methodology: how it impacts each of these components, and how they relate to the variance estimate as a whole. We also address some challenges, and identify areas of potential improvement. Finally, we use data from the CPS to show how the methods are working.
|