Sampling Procedure
For CAS 2021 a new screening activity was conducted. The method for CAS 2021 involved the selection of a representative sample of Enumeration Areas (EA) throughout Cambodia from the 2019 General Population Census Sampling Frame. Households within these EAs were screened for any agricultural activity. Using this basic information, the agricultural households were stratified and sampled for additional data collection. A simultaneous data collection effort focused on juridical holdings, utilizing listings of non-household agricultural operations, expanded based on feedback from MAFF and NIS provincial offices and other stakeholders.
Enumerators along with field and data supervisors were mobilized in all provinces and trained for this undertaking. All enumerators and supervisors (337 enumerators, 46 field supervisors and 46 data supervisors) were existing staff of NIS or MAFF. The collaborative effort between these two ministries was consistent throughout the CAS project, with staff from both organizations contributing from the design of the survey to the data collection and analysis.
The target population comprised the households that were engaged in agriculture, fishery and/or aquaculture. Given their low number of rural villages, the following districts were excluded from the frame: Krong Preah Sihanouk (province Preah Sihanouk), Krong Siem Reab (province Siemreap). Khan Chamkar Mon, Khan Doun Penh, Khan Prampir Meakkakra, Khan Tuol Kouk, Khan Ruessei Kaev, and Khan Chhbar Ampov (province Phnom Penh).
For the CAS 2021, the 2019 General Population Census Sampling Frame was utilized. This frame consisted of around 14,500 villages and 38,000 Enumeration Areas (EAs). For each village, the following information was available: province, district, commune, type (rural/urban), number of EAs and number of households. The target population comprised the households that were engaged in agriculture, fishery and/or aquaculture. Given their low number of rural villages, the following districts were excluded from the frame:
- Province Preah Sihanouk, District Krong Preah Sihanouk
- Province Siemreap, District Krong Siem Reab
- Province Phnom Penh, District Chamkar Mon
- Province Phnom Penh, District Doun Penh
- Province Phnom Penh, District Prampir Meakkakra
- Province Phnom Penh, District Tuol Kouk
- Province Phnom Penh, District Ruessei Kaev
- Province Phnom Penh, District Chhbar Ampov
Since the number of rural households per EA was not known from the 2019 census, to calculate the number of rural households in each province, the sum of the households in the villages that were classified as rural was computed. The listing operation in each sampled EA was conducted for the CAS 2021 to identify the target population, i.e., the households engaged in agricultural activities.
For this survey, there was no minimum threshold set to determine a household's engagement in agricultural activities. This differs from the procedures used during the 2013 Agriculture Census (and that would be used in the 2023 Agriculture Census later), in which households were eligible for the survey if they grew crops on at least 0.03 hectares and/or had a minimum of 2 large livestock and/or 3 small livestock and/or 25 poultry. The procedure used in the CAS, which had no minimum land area or livestock or poultry inventory, allowed for smaller household agricultural holdings to have the potential to be selected for the survey. However, based on the sampling procedure indicated below, household agricultural holdings with larger land areas or more livestock or poultry were identified and associated with different sampling strata to ensure the selection of some of them.
The CAS 2021 used a two-stage stratified sampling procedure, with EAs as primary units and households engaged in agriculture as secondary units. In the CAS 2021, 1,381 EAs and 12 agricultural households for each EA were selected, for a total planned sample size of 16,572 households. The 1,381 EAs were allocated to the provinces (statistical domains) proportionally to the number of rural households. To select the EAs within each province, the villages were ordered by district, commune, and then by type of village (Rural-Urban). Systematic sampling was then performed, with probability proportional to size (number of households). The total effective sample size of the survey was 16,563 agricultural households.