Sample Correlations of Current Population Survey Unemployment Statistics

• Activity Number: 614
• Type: Topic Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Reid Rottach*+
• Companies: U.S. Census Bureau
• Keywords: composite weighting ; calibration ; annual sampling ; phase-in/phase-out ; variance components ; modeling

Abstract

Each month, the Current Population Survey (CPS) interviews housing units (HUs) from eight rotating panels (rotation groups). Each panel is interviewed for four consecutive months, followed by eight months of not being in sample, followed by four more months of interviewing. One of the eight panels is new to the survey in any given month, replacing an outgoing panel selected from the same second-stage cluster of HUs. All panels in all months within a design are selected from the same primary sampling units, which are groups of counties. This approach to sampling results in a complex pattern of correlations among panel totals within and across months. Furthermore, the approach we use to construct estimation weights also affects correlations. We model the effect of different stages of sampling and weighting adjustments on CPS correlations, and in turn, on variances and their decomposition.