We are designing a survey that aims to include both the registered and unregistered drug shops. Though we have the sampling frame for the registered but not for the unregistered ones. In such case how can we design the sampling ensuring the probability method?
It is clear that the problem is only with the unregistered drug stores. As long as there is no statistics on this side, I doubt that you can reach probability sampling. you can do it for registered group, and just mention the total number that you can include and try your best to estimate the possible unregistered drug stores.
A basic norm to design a representative sample is to know the totals of each part of the population. For this, if you do not know the number of unregistered shops it is not possible to have representativity in the sample with respect to the population.
The probabilistic method requires moreover to have identified all the shops as units of the population. This is not the case in your example.
It is clear that the problem is only with the unregistered drug stores. As long as there is no statistics on this side, I doubt that you can reach probability sampling. you can do it for registered group, and just mention the total number that you can include and try your best to estimate the possible unregistered drug stores.
There might be some data on proportion of registered and unregistered drug stores. For each registered drug store how many unregistered stores are there. Although this is an assumption you might get an approximate number of drug stores. So you can take that as your sampling frame. This is my view and I do not have any references to justify this.
Thanks Nipun for your response. However, if available, this will only generate approximate number of unregistered shops. But we also need a list to select the drug store ensuring that every store has equal probability of selection. May be we can state this into our limitation.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey