Leaving aside these biases, think about denominator and numerator issues.
When we estimate newly incident use or dependence, we generally have large denominators to work with. The S.E. help us deal with uncertainty such as the relatively small numbers of users who will die within the first 1-2 years after first use, or soon after dependence onset.
In contrast, the estimates for quitting among cases are constrained by the relatively small number of observed cases in each survey. Wide S.E. as a result, even if we ignore the LT and LC biases.
Make sure you can describe LT versus LC in the survey context.
A paper that I think over-values our epidemiological survey evidence on quitting:
URL for Heyman, 2013Perhaps "clinical investigation" samples are better than epidemiological samples in this domain of inquiry, if they can document zlT and LC bias constraints better than we can?
No comments:
Post a Comment
Comments to this blog are moderated. Urgent or other time-sensitive messages should not be sent via the blog.