I am trying to sample Poultry farmers in a locality for a study. But I don't have information on the number of poultry farms in the area. How then do I obtain the sample size for this study?
Do you have access to national agricultural census? It can help you.
Godwin Dogbey -
The two comments above sound like possibilities for determining the population. Then you still need a sampling technique. Stratified random sampling would do much better than simple random sampling, if you can learn enough about your population to stratify well.
Another possibility, which however does not require a list of farms, would be area cluster sampling where you could consider the "locality" you mentioned, and divide a map of it into equal size grids, select the grids in a simple random sampling, or stratified random sampling, and you could census the farms you find inside the grids you randomly select.
Unequal probability sampling can be good, but more complex, and it would only be ideal for one item on a quantitative survey. The size measure may not be very good for other items/variables of interest.
My area is prediction (meaning regression, not forecasting), which may work well for businesses if you have good predictor/auxiliary data on the population (or in strata/subpopulations), and bias is not a problem. As a simple example for model-based classical ratio estimator prediction, see
https://www.researchgate.net/publication/261947825_Projected_Variance_for_the_Model-based_Classical_Ratio_Estimator_Estimating_Sample_Size_Requirements. This works very well for sample surveys when you have a previous census of the same items. For inference for each item, unlike unequal probability sampling, each item has its own measure of size. You might consider that for a long term plan.
Hope something above is helpful. If you want to know more about some terms, an internet search on such a term along with "Pennsylvania State University" may be useful. They have a great deal of good information online.
Cheers - Jim
- Badges
- Science method