Dear all,
I have 2 cross sectional surveys data (before and after) on 2 group (of villages), group A and group B, and I would like to estimate the DID of a treatment on A compared to B. The survey was conducted on the same villages during both surveys, but not on the same individuals. So I considered a first level of cluster 'village' . The target was children below 2 years of age. During the survey it happened to include 2 kids in the same household (who may share the same characteristics). I would like to take this into account by adding a second level of clustering in the analysis (for example 'household').
I used this syntax in stata 12.1
where 'stunting' is a binary outcome, 'mnp' is the treatment (0=control ; 1= treatment) and 'enquete' the period (0=before ; 1= after)
Question: How to include the second level of clustering (household) in this model? Would the syntax work for continuous outcome?
Alternativley I was thinking of using GLLAMM as follows:
Question: If this possible, what would be the correct syntax and equations for the GLLAMM . Also can I use the same syntax for continuous outcomes?
Thank you very much for your help.
---------
Lanou Bienou
I have 2 cross sectional surveys data (before and after) on 2 group (of villages), group A and group B, and I would like to estimate the DID of a treatment on A compared to B. The survey was conducted on the same villages during both surveys, but not on the same individuals. So I considered a first level of cluster 'village' . The target was children below 2 years of age. During the survey it happened to include 2 kids in the same household (who may share the same characteristics). I would like to take this into account by adding a second level of clustering in the analysis (for example 'household').
I used this syntax in stata 12.1
Code:
regress stunting mnp##enquete, vce(cluster village)
Question: How to include the second level of clustering (household) in this model? Would the syntax work for continuous outcome?
Alternativley I was thinking of using GLLAMM as follows:
Code:
gllamm stunting mnp##enquete, family(poisson) link(log) i(household village) eqs(int) nip(8) adapt eform robust
Thank you very much for your help.
---------
Lanou Bienou
