S-2 had sent me the following links and asked me to comment.
http://news.bbc.co.uk/2/shared/bsp/h...afghanpoll.pdf
ABC News/BBC/ARD Afghanistan Poll ? Note on Methodology - ABC News
Since I haven't followed Afghanistan as closely as I followed Iraq, I can't speak smartly to the trends that may or may not be there. I'd be careful in deriving potential inferences from the way the numbers trend simply because while there is typically five or so years of data, I believe that the timing could have a huge influence. For example, in this poll, a recent heavy snowstorm may result in more negative responses this year, whereas from last year's poll, it was taken at the end of the fall when a good harvest was in progress. While both those examples are hypothetical, it's something to be cautious of when the numerical differences aren't that great.
Second, the methodology for this poll was cluster sampling - the same methodology as the Roberts et al (typically referred to as the Lancet study) on excess deaths in post-war Iraq. Just as I spoke about that study, this doesn't damn it, but it does result in wider confidence intervals, or in survey parlance, a larger margin of error. Their number of total observations is low, and so while I can't translate that to a margin of error, it will be greater than your typical +/-3 that you see for most political polls. In their favor, they are very explicit about having to replace samples and non-response bias, which is okay for the former and very good for the latter.
Thirdly, because the poll language doesn't distinguish Taliban from taliban, I'd be weary of some of the potential inferences to be made about attitudes towards taliban members. Folks may not like the Taliban, but they may support the taliban, and so an answer against the Taliban combined with a thumbs up for an increase in security that is the result of the taliban could be bad news for ISAF unless that taliban element is willing to join the coalition fold. However, a naive interpretation in this case would be that the increased security is the result of ISAF and so it's good news for ISAF. Once again, this is a hypothetical, and since the level of observation in these results is nationwide, there's no way to disentangle whether it's a plus or minus for ISAF.
For S-2, sorry that I'm not able to provide more insight than to provide caution in how to potentially read the survey results, but I don't want to be pulling stuff out of my fourth point of contact :(
http://news.bbc.co.uk/2/shared/bsp/h...afghanpoll.pdf
ABC News/BBC/ARD Afghanistan Poll ? Note on Methodology - ABC News
Since I haven't followed Afghanistan as closely as I followed Iraq, I can't speak smartly to the trends that may or may not be there. I'd be careful in deriving potential inferences from the way the numbers trend simply because while there is typically five or so years of data, I believe that the timing could have a huge influence. For example, in this poll, a recent heavy snowstorm may result in more negative responses this year, whereas from last year's poll, it was taken at the end of the fall when a good harvest was in progress. While both those examples are hypothetical, it's something to be cautious of when the numerical differences aren't that great.
Second, the methodology for this poll was cluster sampling - the same methodology as the Roberts et al (typically referred to as the Lancet study) on excess deaths in post-war Iraq. Just as I spoke about that study, this doesn't damn it, but it does result in wider confidence intervals, or in survey parlance, a larger margin of error. Their number of total observations is low, and so while I can't translate that to a margin of error, it will be greater than your typical +/-3 that you see for most political polls. In their favor, they are very explicit about having to replace samples and non-response bias, which is okay for the former and very good for the latter.
Thirdly, because the poll language doesn't distinguish Taliban from taliban, I'd be weary of some of the potential inferences to be made about attitudes towards taliban members. Folks may not like the Taliban, but they may support the taliban, and so an answer against the Taliban combined with a thumbs up for an increase in security that is the result of the taliban could be bad news for ISAF unless that taliban element is willing to join the coalition fold. However, a naive interpretation in this case would be that the increased security is the result of ISAF and so it's good news for ISAF. Once again, this is a hypothetical, and since the level of observation in these results is nationwide, there's no way to disentangle whether it's a plus or minus for ISAF.
For S-2, sorry that I'm not able to provide more insight than to provide caution in how to potentially read the survey results, but I don't want to be pulling stuff out of my fourth point of contact :(
Comment