Malaria Indicators Survey 2005

The last Population and Housing Census for Liberia was conducted in 1984. At that time, a National Sampling Frame comprising of 4,800 EAs was constructed for the Census. Not only is this frame 20 years old, but there has also been many undocumented changes in the size, structure and distribution of population and dwellings. During the 20 years period, many new communities were established, while existing ones had expanded or contracted due to migration and changes in socio-economic development. Additionally, the civil war led to the destruction of many communities and dwellings and massive displacements of the population of certain communities, as well as the death of thousands of people. These new developments have not been documented, what makes the existing sampling frame to be outdated and obsolete to be used in selecting samples for the Liberia Malaria Indicators Survey (LMIS) without undertaking a major verification exercise.
In 1999/2000, the Government of Liberia and its partners conducted a Demographic and Health Survey (DHS) using the 1984 National Sampling Frame to select and verify 600 EAs. Even though these 600 EAs are more than 5 years old now and were affected by recent civil wars, they have been considered a National Sampling Frame for the LMIS, since in fact they were nationally selected, verified and can be easily identified and verified at this point of time, thereby saving cost.

A two-stage (EAs at first stage and households at second stage) stratified sample design was developed for the 2005 LMIS. At the first stage, the 600 EAs were stratified by county and rural-urban residence. There are 15 counties and 2 rural-urban strata. The urban stratum comprised all county capitals/headquarters plus the City of Monrovia and was divided into 2 domains: (1) City of Monrovia and (2) the county capitals/headquarters. The rural stratum included the rest of the country, that is, all areas outside of the City of Monrovia and the county capitals/headquarters.
Hence, a total of 17 sampling strata were identified, that is, the City of Monrovia, all county capitals/headquarters grouped together as one stratum, and the reminder (rural parts) of the 15 counties. In view of the foregoing and based on the availability of funds, a total of 360 EAs (60%) were selected out of the 600 EAs with probability proportional to size, that is, to the number of households/structures in 1999/2000 DHS Sampling Frame.
In addition to identification and updating of EAs, a complete listing of dwellings/households in the EAs was necessary prior to the selection of households. The listing operation consisted of visiting each of the selected clusters, recording on listing forms a description of every structure together with the names of the heads of the households found in the structure, and drawing a location map of the cluster as well as a sketch map of the structures in the cluster. Twenty field workers (4 teams of 8 interviewers, 4 cartographers and 4 supervisors) completed household listing exercises of the 360 EAs in approximately 40 days (February 25-first week in April, 2005). At the end of the exercise a total of 9000 households were statistically selected for the LMIS from a total of 29,198 households

The selection of EAs within each stratum was undertaken through the following steps:

i) Assign measures of size (MoS) to sample EAs based on the 1999/2000 number of households/structures listed. Let the MoS for the ith EA in the hth stratum = Mhi
ii) Cumulate the Mhi values, i.e., = Mh
iii) Compute the sampling interval (I) as given in the formula below:
Where:
ah = assigned number of sample EAs (see column 3, table 1) for the hth strata,

iv) Using a random number table, find a random number, R, between one and I.
v) Compute the sequence of sampling numbers:
R; R+I; R+2I; R+3I; etc.

B. Cluster Selection Probabilities:
The cluster selection probability for the ith cluster in the hth stratum will be given by:
where: fhi = 1st stage selection probability for the ith PSU in the hth stratum

Since the EAs are sub-samples of the LDHS, the actual selection probability for the ith EA will be given by:
Where phi = LDHS selection probability for the ith Cluster in the hth stratum

See more on sampling in the final report.