You can find it here (you'll need RealPlayer or RealAlternative to view it). Here is a description of the study, conducted by myself, Brendan Maughan-Brown (University of Cape Town), Nicoli Nattrass (University of Cape Town) and Jennifer Ruger (Yale) from a previous post.
We are currently working on a new draft of this paper and should have that ready in a week or so. In the meantime, I'd love to hear your comments on how to improve upon this study.
6 comments:
Hi there,
I work as an HIV and AIDS journalist in Joburg and would love to write something up on the study when you've finished the draft - it's an issue we've been hearing rumours about for years and it's great to see someone's looked into it. Please give me a shout at laura(at)plusnews.org.
-Laura
I watched the whole thing. I'm going to have to take a look at the literature you describe that defends the general health status measures. I'm definitely going to cite some of those papers in my work.
Less academic that the prior posts-Atheen, didn't know you worked with Natrass! Very cool, I've really enjoyed several of her prior papers.
Although I am hesitant to bring up a more academic point about this, I am wondering what the IV estimators ended up doing in this sample. I'm running some IV specifications on a sample of about 900 now and wondering if it makes sense.
It ended up that we were really flogging the data with IV since we had 240 observations in a t=3 panel with individual fixed effects... hence we only use it as a "specification check"...in the final version I think we are going to stay away from the IV stuff completely.
It would be fun to work more on that IV application and do some bootstrapping and maybe even monte carlo simulations. But theres probably more important things to do unless you're interested in publishing in a statistics journal.
Post a Comment