The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Ethiopia » Multilevel modeling
Multilevel modeling [message #25564] Wed, 09 November 2022 13:53 Go to next message
gebretsh@gmail.com is currently offline  gebretsh@gmail.com
Messages: 17
Registered: June 2022
Member
Dear DHS experts,

I would like to create a cluster variable using individual level factors (education*wealth*residence) and use it as a sampling unit
in my multilevel modeling. The PSU (enumeration area) and SSU (households) are already available in the data.

1) could I use PSU, SSU and this newly created cluster as the third sampling stage in three level multilevel model?

2) could I use PSU and this newly created cluster as the second stage in a two level multilevel model?

3) could I use HH as PSU and this newly created cluster as the SSU in a two level multilevel model?

My aim is to study inequality/variation (via ML model) in health care use across the new cluster variable.

I appreciate in advance for the technical support
Regards,

[Updated on: Wed, 09 November 2022 13:56]

Report message to a moderator

Re: Multilevel modeling [message #25574 is a reply to message #25564] Thu, 10 November 2022 07:51 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3172
Registered: February 2013
Senior Member

Following is a response from DHS staff member, Tom Pullum:

This question goes beyond the usual scope of forum questions. Briefly, the main purpose of multilevel models is to correct the standard errors for design effects in multi-stage sampling. Are you trying to impose on the data a design that is different from the one that was actually used? It's my guess that either (a) the model you describe would not run, because the units are not hierarchical, or (b) the results (effects) would be almost the same with the model as without the model. Perhaps other users can suggest something.
Previous Topic: Ethnicity values across surveys
Next Topic: Timing of variables' collection
Goto Forum:
  


Current Time: Sat Oct 12 16:07:50 Coordinated Universal Time 2024