6.1.3. Recommended approach for random sampling

In view of the discussion above, stratified sampling is recommended to achieve good spatial coverage of all main geographical areas in a country or region whose beekeepers are to be surveyed, and to give a more representative sample and more precise estimation. Using proportional allocation (see section 9.) is the simplest way to implement this. By ensuring that the chosen strata are represented in the sample, comparisons can be made, and these stratification factors can also be used as risk factors in modelling the risk of colony loss, for example. Achieving good spatial coverage of the population is also essential for spatial or spatial-temporal model fitting, which requires a high degree of data resolution.