Defining the Master Sampling Frame for agricultural statistics –
1.2. CORE CROP ITEMS
The Global Strategy identifies eight core crop items – including wheat, maize, and rice – that account for major food supplies, a large proportion of land use, and value added to the economy’s GDP. Each country must identify these and other items important to their economy as core data items. The data required for these core items include area planted and harvested, yield, and production. In some cases, the data requirements include early season forecasts of production and final end-of-season estimates of area harvested and amounts produced. Other core items include agricultural inputs, such as use of fertilizer, improved seeds, water, etc. The scope of core data items also comprises periodic information on farms’ economic situation, costs of production, and changes in structure. In some countries, aquaculture products are also included. The populations that may constitute these core items include:
• A list of the names of farm operators or of agricultural holdings. This can be a register formed with information from a recent agricultural or population census or from administrative sources within the country. It is assumed that the sum of the land area operated by each farm operator adds up to the population total for all farmland; and that the sum of the areas planted and harvested for each crop adds up to those population totals. In this case, the sampling unit is the farm operator, the reporting unit is the farm operator or holding and the item of interest is the land operated by the farm operator and all other data variables associated with the holding.
The sampling frame is the list of farms- or landholders and associated data, depending on the source of the list. The data can be information provided by the most recent census or administrative sources. Thisis a list frame.
• A list of rural households. The sampling unit is the household, the reporting unit is the agricultural holding associated with the household and the item of interest is the land operated by the holding, including the area planted and harvested for each crop on the land. It is assumed that the aggregation of the land and crop areas is equivalent to the population totals for the country. In most developing countries, there is one-to-one correspondence between the farm, the farm operator, and the household. An exception is posed by large commercial agricultural enterprises, which generally have business names. The sampling frame is the list of
households and associated data depending on the source of the list. This is a list frame.
• A list of census enumeration areas or small administrative areas such as villages and associated land. The scheme is a two-stage or multiple-stage sample design. The sample units of the first stage are referred to as primary sampling units (PSUs). A sample of PSUs can be selected in a number of ways, such as by stratification and/or PPS sampling. For the first stage of sampling, the frame consists of a complete listing of the enumeration or administrative areas as PSUs. Accompanying information – such as the population, the number of farms, and land areas – can be used for sampling purposes. If the enumeration or administrative areas are geo-referenced, quantitative land cover indicators can be derived from satellite imagery or aerial photographs and used as a sampling tool. The first stage consists inthe selection of a sample of census enumeration or administrative areas as PSUs. The selected PSUs are screened to identify or update the names of farm operators and their linkage to a household. The PSUs can also be screened for rural non-farm households. Within each sample PSU, a subsample is then selected from these listings. At this stage, the sampling unit is a name or a household address, the reporting unit is the agricultural holding associated with the household and the items of interest are the land, crops, livestock, etc. associated with the holding. This subsample can be based on stratification and/or PPS sampling methods (which will be described further detail in later chapters). The sampling frame is the list
of names of farms or the list of households and associated data depending on the source of the lists. However,
these are list frames. The linkage to the selected PSU must be maintained for estimation purposes. The frame
is complete if all PSUs have a chance of being selected and the listings within the selected PSUs are complete.
• Area sampling frames. The sampling units are territorial elements and it is not necessary to build an explicit list of units. If crop area and yield are directly observed, the knowledge of the boundaries of the region of interest is sufficientto ensure that the sampling frame is complete. However, if the reporting units are households or farms, area frames must usually be combined with list frames of large farms or farms producing rare items.
• Segments with natural (or physical) boundaries. These are usually sampled in two steps. First, the territory is divided into blocks that are larger than the intended size of sample segments. These blocks are usually called PSUs and can be stratified by type of land cover and can be sampled. The selected PSUs are divided into smaller
units (segments) and one or more of them are sampled. If only one PSU is selected, it is difficult to estimate the sampling errors unless replicated sampling is used as described in Davies (2009). The reporting unit can be the land segment or the farms or households that can be linked to the segment (see Chapter 6 for further details). Crop areas can be directly measured by observation on the ground. Yields may be measured on a small sample of points inside the segment (crop cutting experiment). In this case, there is a proper two-stage sampling process, in which the segment is the PSU.
• Segments defined by a geometric grid (usually a square grid). The sampling concepts are the same as those that apply tosegments with physical boundaries.
• Points. In area sampling frames, points are usually not considered as dimensionless geometric units: a certain size (for example, 3 m) is attributed to them for the application of the observation rules in the field. Points can be considered as small segments that contain a single land cover type, except in the case of mixed crops. The reporting unit can be the point, but it can also be the household or farm that operates the field in which the point falls. Points can also be sampled within EAs or small administrative units (PSUs). In this case, a mixed two-stage sampling frame would exist: one list frame of small administrative units, and an area frame therein.
Other core crop data requirements include producer and consumer prices, and early warning indicators of conditions adversely affecting crop production.While these items are important, they are beyond the scope of the MSF. In some countries, production from household plots constitutes a significant part of the nation’s production. Each country will need to determine the scope and coverage of household plots when determining the choice of sampling frame.