Assessing discontinuities and rotation group bias in rotating panel designs

Rotation group bias (RGB) and discontinuities are in fact similar phenomena – both are systematic differences in non-sampling errors in a survey. Therefore, they can be dealt with using common approaches, and in this chapter, the authors review the methods for planning a transition to a new approach, and estimating and adjusting for the changes in the evolution of the series. The authors also review the available methods to quantify discontinuities induced by redesigns of the survey process. They show how time series models can be fitted to the outputs from rotating panels to account for RGB. The authors extend these times series models to estimate discontinuities. They also present some example applications. There are clear indications that the RGB is time dependent. This suggests that the common assumption that RGB can be ignored since it does not affect period-to-period change under a balanced rotation pattern is questionable.