Sri Lanka Survey of Household Economic Activities - 1984-85

Sample Design: A national sample of approximately 24,000 households were used for this survey. A sample 'of this size is required, to obtain reasonably accurate estimates at national' level by sectors and at district level.

The sampling design 'was a stratified two stage design. The lists 'of Census blocks (approximately 80' census units in the Urban Sector and 60 census units, in the Rural Sector) created for the Census of Population and Housing - 1981 were used as the basic frame for selection of primary sampling units. The total number of households, with at least one own- account worker, or at least one self-employed farmer or at least one from each category in each census block was obtained on the computer, with the identification information.

The census blocks having at least; 5. such households at the time of the census were considered as eligible ,primary sampling units (PSUs). Census blocks with less than 5 such households were combined with the adjoining blocks and formed combined blocks before selecting sample' PSUs. The households with at least one member who carry on agricultural or non-agricultural activity on an own-account basis were considered as the secondary sampling units, (SSUs). The SSUs were reached through a first stage sample of approximately 2,400 eligible census blocks.

Since the total number of agricultural and non-agricultural workers would be largely dominated by the agricultural component and since, the focus of the survey was equally on non-agricultural activities, the latter was given a special weight-age. It was done by taking two independent allocations of the first stage sample of 2,400 blocks:
One, on the basis of the number of own-account workers in the agricultural activities, and the other on the basis of the number of own-account workers, in non-agricultural activities and' a simple average of the two was taken as the basis for the' allocation, details of which is given in Appendix III of the Final Report attached in the external resources section. The number so arrived at were rounded up into multiples of 12 so as to facilitate uniform spread of the sample in each district over the 12 months of the survey.

As it was necessary to provide estimates at district level, the allocation thus arrived at had to be adjusted to some extent by increasing the sample size for the smaller districts. As such in the case of five districts, in which the allocated sample size was less than 60 census blocks each, the sample was increased by adding 24 blocks in each case. On the other hand the maximum sample size was limited to 180 blocks, as this number of blocks would be adequate for any district to give reasonably accurate estimates at district level. The total sample size thus worked out to be 2,436 census blocks.

The whole island was stratified into 24 districts and each district into two sub-strata, urban and rural (rural ,includes, the estate sector as well),thus forming 48 strata for the whole island, so that each strata would be more or less homogenous. The sample of blocks assigned to each district was allocated to the urban and rural sectors in proportion to the total number of own account workers in agricultural and non-agricultural activities. An allocation similar to that adopted to the district is not essential for scrotal allocations since non-agricultural activities which dominate the urban sector will in any case get adequate representation in the urban sector and there is no specific interest of agricultural activities in urban sector.
The survey operations were staggered uniformly over the survey period of 12 months, taking one-twelfth of the sample each month. Thus at national level 203 blocks were covered each month. In each district one-twelfth of the sample assigned was taken each month; including approximately one twelfth of the urban sample wherever feasible as shown in Appendix IV of the Final Report attached in the external resources section..

Whenever the number of urban blocks to be surveyed was small, the urban sample blocks were interspersed so that they would spread uniformly over the year. The selection of PSUs · (i.e. the Census Blocks) was done with. Probability Proportional to size (PPS) sampling with replacement, the size being the total number of households with at least one self-employed farmer (a) or with, at least one own-account worker (b) or with at least one self-employed farmer and, at least one own-account worker (c) i.e. the size measure was (a+b+c).

The lists of census units prepared for the Census of Population and Housing 1981, of each selected block were updated to include new housing 'units and to exclude ones which are no longer in existence. This updating operation was also staggered over a period of twelve months
starting from February 1984 to January 1985. The lists in respect of the census blocks which were surveyed in April, were updated in February. The blocks which were surveyed in May, were updated in March and so on. This way it was possible to obtain the most up-to-date lists for the selection of SSUs. While up-dating the list of census units in the selected census blocks, all households within each selected block, were listed on a special listing from (EALF form which is given in Appendix I) and among them households having at least one member who is a self-employed farmer or an own-account non-agricultural worker was identified. The listing schedule has two screening questions: .
(i) Is anybody in the household self-employed in Agriculture either as a primary or secondary activity? if the answer is 'No' to this question, then the following question is asked.
(ii) Is anybody in the household engaged in non-agricultural activity as an own-account worker, either as a primary or secondary activity? '

A positive reply to either of these questions implies that the household is eligible for selection. After identifying the eligible households they were classified by their main household economic activity. The main household economic activity of the household was decided on the· income criterion. It should be noted that the main household economic activity refers only to the activities that are being carried out on household basis and not the activities carried out by the household' members who are only employees working. for others. Thus, if the household is getting more income from agricultural activities carried out on household basis, such households were considered as mainly agricultural households. Similarly, if the household gets more income from non-agricultural activities carried out on household basis, such households were considered as mainly non-agricultural households. The agricultural and non-agricultural households were then sub classified broadly by income level for purposes of sample selection.

This was done by classifying the households in the two groups by-monthly income as "above .
Rs. 800" and "Less than or equal to Rs. 800". For selection of sample households in the rural blocks the eligible households were arranged in the following manner,.
(i) Agricultural Household with a monthly income of above Rs. 800/-.
(ii). Agricultural Household with a '''marginal and unknown" (Le. if the household members 'are uncertain) monthly income..
(ill) Agricultural Household with a monthly income of less than or equal to Rs. 800/-
(iv) Non-agricultural Household with a monthly income of above Rs. 800/-.
(v) Non-agricultural Household with a "marginal and unknown" (i.e. if the household members are uncertain) monthly income.
(vi) Non-agricultural Household with a monthly income of less than or equal· to Rs. 800/-.

In the case of rural blocks the selection of eligible households were done. separately in respect of the' agricultural and non-agricultural groups. Among the Agricultural Households, a sample of'5 households were selected by systematic sampling. Similarly, another 5· households· were selected among the non-agricultural households. If the number happens to be less than 5 .in any group, all eligible households in that group were included in the sample and increased the number to be selected from the other group so that the total number of samples per block is 10 households If both groups did not have the required number of eligible households, all the eligible households in the block were included in the sample.

In the case of urban blocks 10 eligible households were selected by systematic sampling without considering the agricultural and non-agricultural households as separate strata but maintaining same order of rearrangement of eligible households as explained before for rural blocks. (i.e. (i) to (vi) above). If the number of eligible households were less than or equal to 10,. all the eligible households available in the block were included in the sample.

In the rural as well as urban blocks if the number of eligible households in a block was found to be less than 7, an additional sample block was allocated to that particular district in the following monthly sub round, from among the already selected blocks for that district. This was done to maintain the required number of sample households for the district.