Sampling Procedure
The primary objective of the sample design for the 2008 EDHS was to provide estimates of key population and health indicators including fertility and child mortality rates for the country as a whole and for six major administrative regions ( Urban Governorates, urban Lower Egypt, rural Lower Egypt, urban Upper Egypt, rural Upper Egypt, and the Frontier Governorates). In the Urban Governorates, Lower Egypt, and Upper Egypt, the 2008 EDHS design allowed for governorate-level estimates of most of the key variables, with the exception of the fertility and mortality rates. In the Frontier Governorates, the sample size was not sufficiently large to provide separate estimates for the individual governorates. To meet the survey objectives, the number of households selected in the 2008 EDHS sample from each governorate was not proportional to the size of the population in the governorate. As a result, the 2008 EDHS sample is not self-weighting at the national level, and weights have to be applied to the data to obtain the national-level estimates.
The sample for the 2008 EDHS was selected in three stages. The first stage included selecting the primary sampling units. The units of selection were shiakhas/towns in urban areas and villages in rural areas. A list of these units which was based on the 2006 census was obtained from CAPMAS, and this list was used in selecting the primary sampling units (PSUs). Prior to the selection of the PSUs, the frame was further reviewed to identify any administrative changes that had occurred after the 2006 Census. The updating process included both office work and field visits for a period of around 2 months. After it was completed, urban and rural units were separately stratified by geographical location in a serpentine order from the northwest corner to the southeast corner within each governorate. During this process, shiakhas or villages with a population less than 2,500 were grouped with contiguous shiakhas or villages (usually within the same kism or marquez) to form units with a population of at least 5,000. After the frame was ordered, a total of 610 primary sampling units (275 shiakhas/towns and 335 villages) were selected.
The second stage of selection involved several steps. First, detailed maps of the PSUs chosen during the first stage were obtained and divided into parts of roughly equal population size (about 5,000). In shiakhas/towns or villages with a population of 100,000 or more, three parts were selected, two parts were selected from PSU's with population 20,000 or more (and less than 100,000). In the remaining smaller shiakhas/towns or villages, only one part was selected. Overall, a total of 998 parts were selected from the shiakhas/towns and villages in the 2008 EDHS sample.
A quick count was then carried out to provide an estimate of the number of households in each part. This information was needed to divide each part into standard segments of about 200 households. A group of 48 experienced field workers participated in the quick count operation. They were organized into 15 teams, each consisting of 1 supervisor, 1 cartographer and 1 counter. A one-week training course conducted prior to the quick count included both classroom sessions and two field practices in a shiakha/town and a village not covered in the survey. The quick-count operation took place between the end of October 2007 and end of December 2007.
As a quality control measure, the quick count was repeated in 10 percent of the parts. If the difference between the results of the first and second quick count was less than 2 percent, then the first count was accepted. No major discrepancies were found between the two counts in most of the areas for which the count was repeated.
After the quick count, a total of 1,267 segments were chosen from the parts in each shiakha/ town and village in the 2008 EDHS sample (i.e., two segments were selected from 561 PSUs and three segments from 48 PSUs and one segment from one PSU). A household listing operation was then implemented in each of the selected segments. To conduct this operation, 14 supervisors and 28 listers were organized into 14 teams. Generally, each listing team consisted of a supervisor and two listers. A one-week training course for the listing staff was held at the beginning of January 2008. The training involved classroom lectures and two days of field practice in three urban and rural locations not covered in the survey. The listing operation took place during a six-week period, beginning immediately after the training.
About 10 percent of the segments were relisted. Two criteria were used to select segments for relisting. First, segments were relisted when the number of households in the listing differed markedly from that expected according to the quick count information. Second, a number of segments were randomly selected to be relisted as an additional quality control test. Overall, the discrepancies found in comparisons of the listings were not major.
The third stage involved selecting the household sample. Using the household listing for each segment, a systematic random sample of households was selected for the 2008 EDHS sample. All evermarried women 15-49 who were present in the sampled households on the night before the survey team visited were eligible for the main DHS interview. In addition, in a subsample of one-quarter of the households in each segment, all women and men age 15-59 who were present in the household on the night before the interview were eligible for the health issues interviews and the hepatitis C testing.
Note: See detailed description of the sample design in Appendix B of the survey report.