An official website of the United States government
There are many papers discussing the estimation of variances for seasonally adjusted data. Serious research extends back to at least the early 1980s, yet there are few instances of official adoption by statistical agencies. Although there are several proposed methods to do so, there are no options in widely used seasonal adjustment software packages. Because of this, users often apply variance measures estimated with not seasonally adjusted data to seasonally adjusted data and assume there are no differences. We apply two methods to estimate variances of seasonally adjusted national Current Population Survey series and compare the results. The first method uses replication variances for seasonally adjusted estimates and the second incorporates linear filter weights based on the seasonal adjustment model.