When non-responses are collected, all units need to be declared alongside a reason for non-response.
In the commercial fisheries world we are in a situation where we have concrete data on non-responses that can be used to evaluate and adjust the estimates. Having the possibility of providing that additional information (e.g., details of the fishing activity of non-responding vessels) is important evaluate biases in our sampling and correct (and check) our estimation.
To add that info one of two alternatives need to be possible
1. we declare non-responses as XXsampled==N and are able to upload some lower data when we have it.
2. we declare non-responses as XXsampled==Y and do not add lower data when we don't have them.
Either one works. Option 1 seems better because it provides for simple filtering of what was actually sampled: if VSsampled==N one deletes the lower branches before the counts.
When non-responses are collected, all units need to be declared alongside a reason for non-response.
In the commercial fisheries world we are in a situation where we have concrete data on non-responses that can be used to evaluate and adjust the estimates. Having the possibility of providing that additional information (e.g., details of the fishing activity of non-responding vessels) is important evaluate biases in our sampling and correct (and check) our estimation.
To add that info one of two alternatives need to be possible
Either one works. Option 1 seems better because it provides for simple filtering of what was actually sampled: if VSsampled==N one deletes the lower branches before the counts.