690 – Analyzing the American Community Survey
Investigating Methods to Support Subannual Estimates in the American Community Survey
Keith A. Albright
U.S. Census Bureau
Mark E. Asiala
U.S. Census Bureau
The Census Bureau uses American Community Survey (ACS) data to produce, on an annual basis, one-year, three-year, and five-year estimates, for time periods defined by calendar years. The data for these estimates are collected in yearly samples that are equally distributed across monthly panels, but monthly response patterns are not equal. Numerous external and internal stakeholders have expressed interest in estimates that are based on shorter subannual time periods, such as monthly estimates. The variable monthly response patterns mean that annualized weights, from the pooled samples, cannot be used to produce subannual estimates. This paper presents results of a pilot project that attempts to optimize the existing weighting methodology to produce monthly estimates of health insurance coverage using data from the monthly ACS samples. If successful, these methods could be used to study trends in fast changing characteristics or possible seasonal patterns in the data in a way that is not currently possible with annualized weights.