Egypt, Arab Rep. - Household Income, Expenditure and Consumption Survey 2008-2009

The sample of HIECS, 2008-2009 is a two-stage stratified cluster sample, approximately self-weighted, of nearly 48000 households. The main elements of the sampling design are described in the following.

Sample Size
It has been deemed important to retain the same sample size of the previous two HIECS rounds. Thus, a sample of about 48000 households has been considered. The justification of maintaining the sample size at this level is to have estimates with levels of precision similar to those of the previous two rounds: therefore trend analysis with the previous two surveys will not be distorted by substantial changes in sampling errors from round to another. In addition, this relatively large national sample implies proportional samples of reasonable sizes for smaller governorates. Nonetheless, over-sampling has been introduced to raise the sample size of small governorates to about 1000 households As a result, reasonably precise estimates could be extracted for those governorates. The over-sampling has resulted in a slight increase in the national sample to 48658 households.

Cluster size
An important lesson learned from the previous two HIECS rounds is that the cluster size applied in both surveys is found to be too large to yield an accepted design effect estimates. The cluster size was 40 households in the 2004-2005 round, descending from 80 households in the 1999-2000 round. The estimates of the design effect (deft) for most survey measures of the latest round were extraordinary large. As a result, it has been decided to decrease the cluster size to only 19 households (20 households in urban governorates to account for anticipated non-response in those governorates: in view of past experience non-response is almost nil in rural governorates).

A more detailed description of the different sampling stages and allocation of sample across governorates is provided in the Methodology document available among the documentation materials published in both Arabic and English.


For the total sample, the response rate was 96.3% (93.95% in urban areas and 98.4% in rural areas).

Response rates on the governorate level at each sampling stage are presented in the methodology document attached to the documentation materials published in both Arabic and English.


In order for the sample estimates for the HIECS to be representative of the population, it is necessary to multiply the data by a sampling weight, or expansion factor. The basic weight for each sample household would be equal to the inverse of its probability of selection (calculated by multiplying the probabilities at each sampling stage). The HIECS sample is approximately self weighting at national level and strictly self-weighting at the governorate level, it should be easy to attach a weight to each sample household record in the computer files, and the tabulation programs can weight the data automatically. The sampling probabilities at each stage of selection will be maintained in an Excel spreadsheet so that the overall probability and corresponding weight can be calculated for each sample cluster.

The procedures for calculating the weights and variances are described in details in the methodology technical document attached to the documentation materials published in both Arabic and English.