Incorporating Variance and Geographic Specificity into the Imputation Frame Used in Weighting the American Community Survey Group Quarters Sample

• Activity Number: 242 - Issues in Frame Quality and Accuracy Assessments
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Dirk Bullock* and John M. Jordan and Edward C. Castro, Jr.
• Companies: U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau
• Keywords: Group Quarters; American Community Survey; Estimation; Variance Estimation; Weighting

Abstract

The American Community Survey, Group Quarters (GQ) imputation frame is the source of information about each GQ eligible for imputation, including the population. Each year, the population is accurately updated in sampled GQs. However, the population is uncertain for GQs that are not sampled. Currently, this population is estimated with updates from sampled GQs. The populations are updated using a national level adjustment and do not currently account for the sampling variance. The purpose of this research is to look at updating GQ populations at lower levels of geography and incorporating the sampling variance into the frame. The updates are calculated at the state level and collapsed to higher levels of geography to meet sample size and extreme adjustment criteria. Variance is incorporated into this adjustment through the use of replicate weights, which enable it to easily be incorporated with the GQ estimation procedure. By including both a collapsing algorithm and variance in the imputation frame, the variance of the GQ estimates are improved.