Quality procedures for survey transitions - experiments, time series and discontinuities

Abstract

To maintain uninterrupted time series, surveys conducted by national statistical institutes are often kept unchanged as long as possible. When a change is proposed to improve the methods, it may affect the continuity of these series. It is important to minimise the impact so as to minimise the inconvenience for users. In this paper we set out the steps in an orderly transition, provide practical guidance on how to minimise discontinuities, and review methods for dealing with discontinuities if they arise so as to maintain a consistently-estimated series.