Home » Data » Weighting data » De-normalizing weights and svyset command in Stata
| De-normalizing weights and svyset command in Stata [message #3570] |
Mon, 12 January 2015 08:48  |
hannekeyserhegdahl
Messages: 9
Registered: October 2014
|
Member |
|
|
Hi!
I have several questions regarding pooled datasets, weighting and svyset-command in Stata:
1) I am pooling datasets from multiple countries into regions to estimate HIV prevalence ratios between men and women in specific regions and to compare ratios from different regions. I have understood that I should de-normalize the weights in each of the countries before I pool them together, however I do not understand exactly how this is done. In the formula for de-normalizing weights, which value for v005 do I use/how do I find this value?
2) Should I do anything more to these de-normalized weights for Stata to understand that they represent different countries (like with the strata and cluster variables)?
3) I have already appended male and individual recode files and merged this "pooled" file with the HIV file for each of the countries. To asses HIV prevalence ratio, I am using the hiv05 (weight from HIV dataset). I have also done analyses on these datasets (each country separately) and did nothing to the hiv05-weigth, however, now I am a bit confused as to wether I should do something to this weight or not, since these datasets also are pooled in a way..
4) I have also understood that I should always use the svyset/svy command in analysing survey data in Stata, but I am using "GLM for the binomial family: binreg" (binreg is not supported by the svyset command), is it possible to work around this in some way?
5) When creating country-specific cluster and strata variables, to what value do I add the different 10000's (10000, 20000, etc), or do I just rename the variables?
Hope someone can clear up these questions for me, as I don't understand everything being discussed in previous threads.
|
|
|
|
Current Time: Sat Nov 30 21:26:12 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
