A comparison of six sampling schemes for price index construction in a COICOP food group

Abstract

This article compares the behaviour of sampling techniques for price indices using a scanner data set as a model population. Indices produced by two purposive deterministic cut-off designs and four probabilistic sampling schemes are compared with each other and with the ‘true’ population index from the whole data set. We found that the two deterministic cut-off sampling schemes show much different behaviour from the probabilistic sampling schemes. This is not unexpected, as the former schemes have a very restricted focus with respect to the variety of products. We also found that the probabilistic schemes are generally closer to each other and the ‘true’ value than the deterministic cut-off designs. The jackknife resampling technique is also explored as a means of estimating the SE of the index and compared with the actual results from repeated sampling.