Use of Administrative Data To Explore Effect of Establishment Nonresponse Adjustment on the National Compensation Survey Estimates

Chester H. Ponikowski and Erin E. McNulty


Non-response is a common but undesirable feature of a survey. It may lead to biases in survey estimates and an increase in survey sampling variance. Survey practitioners use various techniques to reduce bias due to non-response. The most common technique is to adjust sampling weights of responding units to account for non-responding units within a specified set of weighting classes or cells. In the National Compensation Survey (NCS) the weighting cells are formed using available auxiliary information: ownership, industry, and establishment employment size. In this paper, we explore how effective the formed cells are in reducing potential bias in the NCS estimates. We use administrative data to estimate average wages for responding units in the NCS. We generate and compare full sample wage estimates to estimates based on responding units with weights adjusted for non-responding units.