Online Program

Sampling Schemes for Price Index Construction: A performance comparison across the COICOP food groups

Keywords: price indices, pps, repeated sampling, jackknife resampling

Price indices are obtained by sampling from a homescan market research data set for the 13 individual COICOP food groups across 21 months. Six sampling schemes – two cut-off and four probability-proportional-to-size (pps) methods – are employed and the true population values used as a reference to derive error measures such as RMSE, average bias and standard deviation. Repeated sampling is employed to estimate the performance of the pps schemes and regression analysis is used to examine how the different schemes, and other variables such as sample size and food group, contribute to these error measures. The use of the jackknife resampling technique for estimating the variance of these indices is also explored.