Estimation of Autocorrelations of Survey Errors with Application to Trend Estimation in Small Areas.

This article describes a simple method of estimating the autocorrelations of survey errors for the case of routing-panel sampling designs. Such designs are in routine use, for example, in labor force surveys (LFS). The survey-error autocorrelations induce spurious trends that need to be separated from the trend of the population values when the latter is of interest. This can be achieved by use of simple state-space models that combine the model holding for the population values with the model implied for the survey errors by the estimated autocorrelations. The procedure is illustrated using data from the Australian LFS.