Variance Estimation in the Swedish Consumer Price Index

Abstract

In large parts of the Swedish Consumer Price Index, products and outlets are sampled independently, yielding a two-dimensional, cross-classified sample. In this article, a design-based variance formula is derived by exploiting the general theory for cross-classified sampling. Numerical variance estimates are given, along with a description of some of the detailed procedures involved in their computation. Finally, the implications of these estimates on the actual allocation are discussed.

KEY WORDS:

• Cross-classified sampling
• Doubly aligned sampling
• Quasi-randomization
• Sampling error
• Two-dimensional sampling