6.4. Sources of sampling frames appropriate for different target populations
The ideal situation is one in which all beekeepers of a country (or other geographical unit) are equally represented in a sample. In some countries, beekeepers, usually with a minimum number of colonies, may be required to register on an official list, in which case gaining access to that list enables access to a selected part of the beekeeping population. In practice, not all beekeepers will register even if this is legally required. The level of compliance with registration requirements may vary greatly from one country to another. In the absence of a satisfactory list of registered beekeepers, other sources of sampling frames may be membership lists of beekeepers’ associations or records held by veterinary services. In some countries (e.g. Norway, Sweden, Denmark and the Netherlands), beekeepers’ associations may represent up to 90% of the beekeepers, so use of these association records seems to be the best approach currently.
Use of any of the above sampling frames for random sampling does require prior consent, by some means, of the beekeepers on the list sampled from, for their record to be used in the selection of a survey sample. Those who would not wish this must have the opportunity to opt out and, having done so, should be omitted from the list before random selection takes place. Ethical approval may also be required (section 4.3.). If cooperation with beekeeper associations which represent the majority of beekeepers in a country is not possible, or complicated because there are many small ones all with a limited number of members, another approach is advised.
Firstly, generally most of the bee stocks in the country are managed by large scale commercial beekeepers (even though there are also large numbers of small scale beekeepers). Often the commercial beekeepers have their own trade organisation which will list them all, as well as the approximate sizes of their operations, and if access can be obtained to them, estimation is possible.
If cooperation with a commercial beekeeper association is not possible, an approach using the fragmented smaller organisations may result at least in some kind of sampling frame from which a sample can be drawn with some hope of being representative of beekeepers who belong to these associations, at least in some local areas. How representative these associations are of all beekeepers is of course unknown.
If no beekeeper infra-structure is available, even if, say, in some parts of the world the post office is the only main central information hub, it may be possible to find for each post office district a nucleus of beekeepers. A representative sample of post offices with respect to climate and suitability of area for beekeeping could be drawn up. If such a sample were not too large, then putting out an enumerator for the survey into each of those post office areas might enable that enumerator to find within that area a fairly complete list of the beekeepers in the area. Then the cluster sampling approach would be sensible, where the post offices sampled were regarded as the clusters. A return would be made for each sampled post office area, and the usual techniques for cluster sampling could be used to analyse the results. These approaches may provide a way forward in situations where there is very little by way of an existing sampling frame and limited resources are available or only small scale surveys are possible. Even an imperfect investigation will yield some information. The important thing in considering the results of such work is to be open about the shortcomings of the results, and not to claim more for them than is justified.