| approximate level-1 and level-2 weights mens dataset NFHS-5 [message #25351] |
Tue, 11 October 2022 02:44  |
vbrar4
Messages: 13
Registered: February 2022
|
Member |
|
|
Hello!
I just had a very quick concern. Currently I am working with the male dataset subset of the NFHS-5 and for my analysis I wanted to proceed with a mulitlevel model but, I am having some issues.
After reading over the resource posted by the DHS program called "Multilevel Modeling Using DHS Surveys: A Framework to Approximate Level-Weights (English)" I have been unable to make any logical progression with determining weights due to missing crucial information in the final report.
The document continuousally refers to tables who's equivalent do not exist in the final report of the NFHS-5.
The document wants information regarding the total number of census clusters and the average number of households per cluster by sampling strata but i cannot find this anywhere.
The two variables of concern produced in the document seem to be A_h and M_h.
For a_c_h the following seems to work:
gen a_c_h=.
quietly levelsof mv022, local(1strata)
foreach 1s of local 1strata {
tab mv021 if mv022==`1s', matrow(T)
scalar stemp=rowsof(T)
replace a_c_h=stemp if mv022==`1s'
}
For m_c the following seem to work:
quietly summarize mv001
gen m_c=r(N)
And according to what i can find in the final report M=636,699, S_h=22 and DHSwt would be DHSwt=mv005/1000000
Any support with how to determine the remaining required information would be appreciated.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
