The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Mortality » multilevel analysis
multilevel analysis [message #5557] Sat, 06 June 2015 06:45 Go to next message
labakecoker is currently offline  labakecoker
Messages: 4
Registered: March 2015
Member

Dear DHS team,
I am working on a mortality analysis of under-five children, and trying to explain for variations in rates across regions in a particular country, and also variations in the decline rates between 2 surveys. what stata command can i use to generate my community level variables? i would also like to know what ideal tool and stata syntax to use in determining which level of determinants best explains for these identified variations.
thank you in advance,
labake coker
Re: multilevel analysis [message #5618 is a reply to message #5557] Wed, 17 June 2015 10:16 Go to previous message
Liz-DHS
Messages: 1516
Registered: February 2013
Senior Member
Dear User,
Here is a response from one of our analysis experts, Shireen Assaf:
Quote:
One way you can create community level variables by cluster is to use the bysort command in Stata. For example we can create a binary variable for education level as follows: recode v106 (0/1=0 "noedu/primary") (2/3=1 "secondary+") (8/9=.), gen(edu)
We can then find the proportion with secondary+ education by cluster as follows: bysort v001: egen educlus=mean(edu)
This will create a new variable, educlus, which would be the proportion of respondents with a secondary+ education level in each cluster. You may also do this by region instead of by cluster. The same procedure can be followed with other variables.
For multilevel models in Stata, you have to install stmixed. Here are a presentation and a journal article, but there are many articles on this type of analysis in the literature:
Presentation: http://www.stata.com/meeting/uk13/abstracts/materials/uk13_c rowther.pdf
Article: http://bioline.org.br/request?ep13024
Another Stata command is mestreg but this is only available with Stata 14.

Thank you!
Previous Topic: Infant and child mortality
Next Topic: Selection of Under-five Mortality sample data
Goto Forum:
  


Current Time: Thu Mar 28 15:27:11 Coordinated Universal Time 2024