|Multiple frame agricultural surveys. v. 1: Current surveys based on area and list sampling methods. (FAO Statistical Development Series - 7) (1996)|
|PART III: THE MULTIPLE FRAME SURVEY DESIGN|
|CHAPTER 10 - MULTIPLE FRAME ESTIMATORS AND THE LIST OF SPECIAL HOLDINGS|
This section sets out the criteria for choosing an adequate list of special holdings, complementary to the area sample survey.
Large Scale producers. Holdings which account for a large share of the total production of some item to be estimated always present a special problem for sampling, and specially for area sampling.
For an area sample, livestock and poultry estimates, using both the closed segment and open segment estimators, are usually the most affected by the occurrence of one or more of these "extreme operators" in the sample. This is readily understood by assuming a situation where a holding with 7,000 cattle is picked up in a segment, although there are only very few such holdings in the stratum. Segment expansion factors often are in the range of 75-100, so that each sample segment is, in theory, representative of up to 100 segments in the population. The inclusion of this holding will completely distort the estimates and increase the sampling variance well beyond what was desired or expected. Also, it is very unlikely that there are 75 or more holdings of that same size in the stratum. The effect on crop area estimates can be equally destructive when using the open segment estimator. A closed segment estimator of crop area may not be as seriously affected but estimates could still be seriously in error in the case of certain crops where there is large scale production in a very localized area, such as occurs with rice in some countries.
The development of a multiple frame procedure that employs a list of the large producers is a procedure most often used to improve the area sample estimates for important variables. A list is needed of such large operations for each variable to be estimated. The list can be compiled from a combination of sources such as; the latest agricultural census, tax records, banks, producers' associations, extension agents, and other local agricultural technicians. The list should contain the location of the holding and the name and address of a person (holder, manager or administrator) who can consistently supply information about the holding. Information about crop areas, total land area, and size of herd or flock are needed in order to be able to place the unit at its proper level of importance on the list. Enough information is needed for a proper identification of the holding if it (or a portion thereof) should be encountered in a sample segment. The list should be updated frequently; once a year or before any major survey if surveys are sporadic. The original sources can be provided with a list of those holdings in their area and asked to delete those who are out of business, indicate new holders of old holdings, and add new holdings.
The size of the list of extreme operations is frequently determined by the capacity to enumerate extra holdings in addition to the area segments. A good place to begin is with the classifications by size of holding from the most recent census to determine the number of holdings in the largest classification for each variable to be estimated. If the sum of the holdings in the largest category is manageable, that can set the size. However, if calculations indicate that encountering a holding or holdings of the next largest size group will increase the variance of the estimate beyond the desired limits, that size group must also be included. After the first survey, the length of the list can be determined more accurately.
The largest holdings on the extreme operator list should all be enumerated if possible. Smaller size classifications can be sampled using simple random sampling. Any duplication between the list and the area sample must be eliminated from the area segment data. One way of eliminating this duplication is by checking the names of holders encountered in the area segments against the list. A better way is to outline on the frame material the area operated by each holding of the list. In this way, it can be determined immediately which segments will contain overlap. The area and location of land operated would have to be updated before each survey because the area operated by this type of large holdings can change significantly from year to year. Estimates from the two sources (area sample and special holding list) are added to make the overall estimate for a province or region.
Localized or Specialized Holdings. A list of holders can also be used in conjunction with the area frame when it is desirable to estimate rare or localized production even when holding sizes are not abnormally large; examples are: cut flowers, vegetables, fish farms, fur farms (mink), high technology poultry, horses, etc. A recent census of agriculture may provide a basic list for these specialty items and indicate the area where they are concentrated. Other sources of lists as well as the management and application of a specialized list are the same as those described above for large holdings.
In view of the considerations and examples above, important survey variables, with a skewed distribution in relation to holdings should be selected. For each of these variables, from the census or other sources, as already indicated, a short list of holdings that account for a relatively large percentage of the total known estimate should be selected to form at least part of the list of special holdings. The complementary list of special holdings should ensure the inclusion of a significant percentage of the total estimate for important survey variables, thus improving the precision of the area estimates.
The list of special holdings should be fairly easy to update annually prior to the survey field data collection, since the holdings involved are usually easy to recognize and well known in the region where they are located. Information can be obtained from extension agents, producers' associations, banks, tax records, agricultural censuses and from government agencies that control and purchase production of certain crops. The preparation of such lists should include the accumulation of data on each operation such as holding size, crops grown, type of livestock and inventory, etc. for stratification purposes if the list is to be sampled. In addition, in order to avoid duplications, a short list will be much easier to combine with the area sample than a large list.
Finally, it should be noted that the additional work involved in constructing a short list of special holdings and in its combination to the area sample is generally minor but gains can be very significant for the estimation of some important variables.