Survey ID Number
BLZ_2001_CAS_v01_M
Title
Child Activity Survey 2001
Sampling Procedure
Administratively, Belize is divided into six districts, namely Corozal and Orange Walk in the north, Belize to the east, Cayo to the west and Stann Creek and Toledo in the south. Each of these districts has distinct urban and rural demarcation. Overall, about 52% of the households in the country are located in the rural areas. Two districts, namely Cayo (which contains the capital Belmopan City), and Belize (which contains the largest urban centre, Belize City), account for almost half the households in the country (CSO, 2001 a).
For the purpose of the Population and Housing Census, each district is sub-divided into smaller Enumeration Districts (EDs). Each ED has an average size of 144 households (Census 2000). For the sampling design of the CAS, available data from both the 1991 and 2000 Censuses were utilised (Tables 2-3 and 2-4). During the time of the preparation of the sample design, the Census 2000 data were not yet computerised and the only available data for 2000 were for household by district and ED and for population by sex, district and ED.
The survey comprised a two-stage design with the selection of EDs being the first stage: the selection of EDs being proportional to the size of EDs, that is, the number of households at the time of the Census 2000. The second stage was the random selection of a cluster of households from within selected EDs. Each district was treated as a stratum in its own right. However, in each case the selected ED and cluster were such that
f = f1 x f2 where
f1 is the probability of selecting an ED,
f2 is the probability of selecting a cluster, and
f is the probability of selecting a household.
To randomly select the 6,058 households, it was necessary to first obtain a distribution of the population 5 to 17 years by district, based on the 1991 Census (Table 2-3). From the data it was also possible to obtain the average number of persons 5 to 17 years per household by district. The 1991 Census also provided data on household income, which were used for ordering EDs within districts prior to selection.
Data on the number of households in each ED were available from the 2000 Population and Housing Census (Table 2-4). These data, together with data on the average number of persons 5 to 17 years per household from the 1991 Census, were used to allocate the number of households to be selected from each district and the urban and rural areas within each district.
Using an average cluster size of 30 households, which gives approximately 200 clusters to yield the 6,058 households and about the same number of EDs, each district was assigned a number of clusters based on the number of households obtained from the Census 2000. After examining the available income data, a monthly household income of BZ$5002 was used as the criterion for ordering the EDs before selection. The EDs were ranked by proportion of households earning less than BZ$500 per month from the highest to the lowest proportion. This was done within each urban and rural area within each district. Within each district, a number of EDs were systematically selected. The selection interval was determined by the number of households in the district and the number of clusters assigned to the district.
After selecting these EDs, one cluster was then selected from each ED. Dividing the number of households by 30 and rounding off as necessary gave the number of clusters assigned to an ED. However, the sum of clusters from the EDs did not exceed the number of clusters assigned to the district. Because the number of households in an ED was not exactly divisible by 30 and the number of clusters assigned to an ED must be an integer, meant that the cluster sizes varied from 27 to 35 households with an average of about 30 households. For example, an ED with 128 households would have been assigned four clusters, each with an average size of 32 households.
Based on the average number of persons aged 5 to 17 years per household from the 1991 Census data, it was estimated that the number of persons aged 5 to 17 years to be interviewed from the sample would be just over 10,000 (the actual number interviewed for the survey was 7,870 children.). An estimated 20% to 25% employment rate for children 5 to 17 years old should have yielded 2,000 to 2,500 employed persons in the sample (the results from the survey showed 896 employed children). Selection of the sample was made at the CSO using the households from the 2000 Population and Housing Census as the sample frame.