4. STATISTICAL SAMPLING PROCEDURES

The conventional approach to statistical sampling of a production system consists of three main components:

1. Definition of the sampling framework, in which the size and main features of the system to be sampled are exhaustively listed, described and mapped.
2. The frame is used to plan sampling activities of a number of officers following a rigorous plan of activities, such that each one of the main productive components has a known probability of being sampled, so that the results of the survey allow the total production to be estimated over the period of the survey with a known degree of error.
3. One or more quality checks are usually built into the survey, detect and allow compensation for biases in sampling.

Such a three stage sampling approach in fisheries is often carried out over a relatively short period of time by a fairly larger number of sampling officers (see Bazigos for details), and may be updated at regular intervals. This procedure is the one to be recommended if temporary manpower (e.g., student assistants) are available, but imposes problems for many small island states where such manpower is not always available. However, it is useful to consider a little further the components described above, in the fisheries context, before looking for alternative approaches.

^a Port of entry

Figure 1  Example of thematic map showing elements useful in constructing a statistical frame. (Modified from Caddy, 1982). (N.B. Although based on the fisheries infrastructure of Bahamas, this is for demonstration purposes only, and is not intended to represent an up-to-date representation of the current situation)

Figure 2  Illustrating one useful way of displaying statistical information for policy makers and planners. (From: Nova Scotia Fisheries Atlas: Nova Scotia Dept. of Fisheries)

4.1 The Sampling Frame

The design of fisheries statistical surveys for collection of the required items of information demands the existence or construction of some form of framework for data collection. Such a framework can be in the form of a map or chart, or an annotated table of key features of the fishery, and is referred to as the “sampling frame”. Such a frame is the first requirement for any proper sampling design, and dictates the field operations that follow. For simplicity, the “frame” may be considered as a map showing principal points of landing for each of the main fishing grounds. This kind of map is sometimes referred to as an “area frame”, and should include principal routes for transportation of fish between landing sites and internal markets, and to markets outside the country (transshipment points), as well as routes for shipment into the country of fish caught outside the national waters, but landed here. An example of such a map is given in Figure 1.

This map should be updated as new information becomes available, and files should be kept for each of these landing points, listed and categorized as in Table 1. Separate, but cross-reference files should be kept also on the principal fish species landed, and on the main fishing grounds.

The chart and filing system described earlier can be cross-reference, and used as a basis for the construction of the sampling frame required for fisheries statistical surveys based on the sample method.

In most cases, Frame Surveys (FSs) based on the census method are designed and executed for the collection of information on the complete spectrum of inventory characteristics required. The results of the FSs are used, among other things, for the establishment of a proper sampling frame.

4.2 Catch Assessment Surveys (CAS)

For the collection of current items information of the input and output characteristics of the operating fisheries (fishing effort, fish catch) CAS's are designed and executed to allow data collection and sampling in both space and time.

The sampling frame established is used for the selection of the sample sites in the CAS. Specifically, for the selection of the sample of landing points (= primary sampling units, PSU's) the method of stratified sampling is employed i.e., landing points are grouped into categories (= strata) according to their importance. An independent sample of landing points is then selected within the established strata. PSU's could then be selected either with equal or unequal probability (see Bazigos (1974) for details).

Sampling in time is employed for the collection of the items of information required within the sample PSU's. Specifically, a number of sample days is allocated to each sample PSU within the established reference periods (i.e., by month, quarter, etc.). Items of information on the survey characteristics are then obtained from a sample of boat-landings within the sample PSU's on the sample days selected.

4.3 Quality Check Surveys

It is important that some time be set aside, particularly by the supervisor, to check on the errors and biases due to either incomplete or inaccurate recording by the statistics gatherers, to investigate conversion factors, and to check on changes or omissions from the sample design uncovered in the cause of the CAS, and to make the appropriate correction, so that these factors can be allowed for in the later stage of summarizing and analysing the data.