Sampling procedure

The 2006-07 Angola Malaria Indicator Survey (2006-07 AMIS) is based on a representative probability sample of households. The sample provides information on women of reproductive age (15-49) and children under five, specifically for malaria-related indicators. The survey covered the entire country.

The sample was designed to provide estimates with acceptable levels of precision for key malaria-related indicators and for two sub-populations: pregnant women age 15-49 and children under five. The major sample domains for which these estimates are computed are:

1. Angola at a national level,

2. Total urban areas and total rural areas of Angola, and

3. Major malaria epidemiologic regions, defined as:

a) Hyperendemic,

b) Mesoendemic Stable,

c) Mesoendemic Unstable, and

d) Luanda, which was extracted from the Mesoendemic Stable region and represents the capital city.

SAMPLE FRAME

Angola is divided into 18 provinces, and they can be grouped into eight sub-regions (e.g., North, East, and Center) according to factors that make some provinces homogeneous among themselves. Each province is subdivided into municipalities (161 in total); each municipality is subdivided into communes (635 in total); and each commune is classified as either urban or rural. Each urban commune is subdivided into administrative areas called censal sections (CSs). Each rural commune has a list of villages, with estimated populations in each village. Therefore, the list of CSs in each urban commune and the list of villages in each rural commune constitute the sample frames for the 2006-07 AMIS.

STRATIFICATION

The communes were grouped by major regions, by rural and urban location, by sub-regions, and by provinces, in order to find homogeneous sampling units. In addition, within each urban commune, several CSs were grouped together to take advantage of the existence of bairros (sub-districts). These groupings were used to stratify the sample.

SAMPLE SIZE

The sample size for the AMIS was estimated based on the minimum size needed to obtain malaria-related indicators with acceptable levels of precision. The precision levels were calculated for each domain. Since the maximum accepted number of domains for Angola was four (i.e., the three epidemiologic regions plus Luanda), the sample size estimate would have to be multiplied by four.

The key indicator selected for the survey was malaria prevalence. Since little was known about its actual level, an assumption was made about its nationwide level. It was estimated between 25 and 30 percent, and the lower level was selected for increased confidence.

Given an estimated malaria prevalence of 25 percent in each domain, at a relative error level of 15 percent, the sample would require 533 children under age five. This is roughly equivalent to 630 households per domain or about 2,500 households nationwide. However, it was possible that the survey would find other indicators at lower percentage values, e.g., children under five sleeping under a bednet. Also, some indicators would be obtained from a smaller sub-population, such as pregnant women. Therefore, the recommended sample size per domain was 750 households, or 3,000 households nationwide. With a sample of this size, depending on the values found, it was possible that some indicators might not be susceptible to analysis at the domain level, but only at the national and urban-rural levels.

SAMPLE SELECTION

The 2006-07 AMIS sample was selected using a stratified three-stage cluster design providing 120 clusters, 48 in urban and 72 in rural areas. In each urban or rural area in a given domain, clusters were selected systematically with probability proportional to size. The selection was done using the following formulas at different stages.

In the first sampling stage, communes were stratified by urban-rural area and by province in each major domain. Then communes were selected with probability proportional to their estimated population using the following formula: P1i = (30 x mi/ mi)

In each selected commune, the second sampling stage selected clusters (censal sections in urban communes and villages in rural communes) with probability proportional to their estimated population size using the following formula: P2ji = (ai x mji/j mji)

The third stage constituted the final selection of households in a given cluster, using the following formula: P3ji = (c/Lji)

The sampling procedures are fully described in Appendix A of " Angola Malaria Indicator Survey 2006-2007 - Final Report" pp.41-45.