Survey ID Number
Child Activity Survey 2016
As in many other household surveys this survey also uses a multistage stratified area probability sample design. The survey uses a two stage stratified sampling design. At the first stage, Census Blocks (census enumeration areas) are selected as primary sampling units PSUs. The CAS 2016 uses 2500 PSUs. At the second stage, ten housing units are selected from each selected PSU as the secondary sampling units (SSUs). The CAS 2016 uses 25000 SSUs.
At the time of the Census of Population & Housing - 2012, Sri Lanka was divided in to approximately 65,000 census enumeration areas (called census blocks) mainly for the convenience of Census enumeration. Before the Census enumeration, all the building units located in these census blocks are listed and the Census was conducted by using these lists of building units prepared at the listing stage. This completed list of census blocks is used as the sampling frame for the household sample surveys conducted by DCS. Usually, the selected census blocks for the survey are updated by making adjustments; in order to accommodate material changes that took place during the period 2012 to 2014. The updated list of housing units in 2014 is used to select the
housing units for the CAS 2016.
It was proposed that, on completion of the Child Activity Survey, the estimates provided should classify the children in major age groups (5-11 years, 12-14 years, 15-17 years) according to Rural /Urban residency, and also that the estimates should classify the activities broadly by status of school attendance as below;
i.Attending school only
ii.Attending school and engaging in economic activities
iii.Attending school and engaging in housekeeping activities
iv.Attending school, engaging in economic and housekeeping activities
v.Not attending school and engaging in economic activities only
vi.Not attending school and engaging in housekeeping activities only
vii.Not attending school, engaging in economic and housekeeping activities
In order to achieve the objectives, it was decided that the survey should be based on a national sample of 25,000 households2. It should be noted that estimates based on sample surveys are always subject to sampling errors and therefore, it is not possible to obtain estimates at every desired level.
Estimates at lower geographical levels such as Province, District, Sector will be provided when such estimates are feasible subject to the relevant sampling errors.
Primary objective of the stratification is to improve the precision of the survey estimates. Usually, the stratification partitions the units in the population into mutually exclusive and collectively exhaustive subgroups or strata. Separate samples are then drawn from each stratum. In addition, the stratification offers administrative convenience and guarantees the representation of important survey domains and sub-populations.
Sri Lanka has twenty-five administrative Districts and DCS has a complete network with all these Districts through the district statistics offices in each District. In many occasions, separate statistics are required at District level for policymaking and decision making at the district level. Considering these, as in many other household surveys the twenty-five Districts are the first level of stratification. Under each district there are three residential sectors namely Urban, Rural and Estate.As in other countries, Urban and Rural sectors have been identified using their socio-economiccharacteristics. However, Estate is a special residential sector for Sri Lanka, which is related tocountry's formal plantation sector and the people who are living and working in these areas. In agiven district, people living in each of these three sectors could be considered to have homogeneouscharacteristics endemic to those sectors. Therefore, the second stage stratification is the sectors in each district. Since there are twenty-five districts in the survey domain we should have 75 strata altogether. However, there are four districts (namely Mullaitivu, Killinochchi, Monaragala andPollonnaruwa) with no identifiable Urban sector and eight districts (namely Jaffna, Mannar,Vavuniya, Mullaitivu, Killinochchi, Batticaloe, Amparan and Trincomalee) with no Estate sector.
Hence, finally sixty-three strata are identified for sample selection.
Primary Sample Allocation and Selection
Two thousand five hundred (2,500) Primary Sampling Units (PSUs) that is sample of 2,500 census blocks were allocated among the 63 strata defined earlier, proportionately to the number of housing units in each stratum (method of probability proportionate to size).
Secondary sample selection
Housing units are the Secondary Sampling Units (SSUs) of the survey. Ten SSUs were chosen from each selected PSU using systematic sampling method. The random start for the selecting process was generated through a computerized programme developed by the DCS. When the total number of housing units in a selected PSU was fewer than 10, then all housing units in that PSU were selected for the survey.