| Multilevel weights level 2 [message #30320] |
Sun, 03 November 2024 16:54  |
Deidi
Messages: 14
Registered: March 2021
|
Member |
|
|
Dear DHS Forum,
I hope all is well. mr
I'm reaching out because I am working on a multilevel model using the 2015 Colombian dataset (IR & MR), and I am trying to figure out how to create second-level weights.
Specifically, I haven't been able to figure out what values to assign for A_h and M_h using Colombia's final reports (Volume I & 2). I haven't been able to identify a specific table with the distribution of EAs and their average size in number of households by region and type of residence (rural/urban) that correspond with the min and max value of v022(min 1, max 65). Also, in the survey, V022 and V023 do not have labels for which regions the values correspond to. If I tabulate v022 and rural/urban, I observe different values for rural and urban, but I'm not sure if that corresponds to A_h or M_h, or neither. Any guidance would be much appreciated.
Warm regards,
Deidi
|
|
|
|
| Re: Multilevel weights level 2 [message #30323 is a reply to message #30320] |
Mon, 04 November 2024 11:49  |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
The files for the Colombia 2015 survey include another geographic subdivision, "Subregion". You can open the HR file and enter the following lines in Stata:
label list hv024
label list hv025
label list SHSUBREG
gen comb=1
collapse (mean) comb, by(hv022 hv024 hv025 shsubreg)
drop comb
list, table clean noobs
list, table clean nolabel noobs
This will help you to identify the geographic areas in the stratification, although with some remaining ambiguity. I wish DHS could help with the creation of cluster-level weights for specific surveys--actually I wish that DHS could have constructed them and made them available for all surveys. However, we are not able to do this. The sampling statistician who was our main expert on this topic, Mahmoud Elkasabi, left DHS two years ago. Perhaps other users can help.
|
|
|
|
| Re: Multilevel weights level 2 [message #30328 is a reply to message #30323] |
Tue, 05 November 2024 12:24 |
Deidi
Messages: 14
Registered: March 2021
|
Member |
|
|
Thank you Bridgette and Dr. Pullum! This response and code are very helpful. I really appreciate it.
And I second Dr. Pullum's sentiment. Making cluster-level weight variables available for all DHS surveys would be great and a huge help!
Warmly,
Deidi
[Updated on: Tue, 05 November 2024 15:02] by Moderator Report message to a moderator |
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
