The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Need to calculate country weight based on facilities when combining two SPA surveys?
Need to calculate country weight based on facilities when combining two SPA surveys? [message #10397] Mon, 25 July 2016 09:39 Go to next message
masjas
Messages: 4
Registered: July 2016
Location: Amsterdam
Member
Using data from the SPA survey Tanzania and Namibia we are interested in exploring the association between health care providers characteristics and their response to question w167 (dichotomized in yes and no).
We want to conduct multilevel mixed effect logistic regression analyses (MELOGIT in STATA), including random effects for health providers, health facility and country. Within each country we are using health provider weights and facility weights. In Kenya 695 facilities are included and in Tanzania 611. I am doubting whether to calculate a country weight to ensure more equal weight of the countries in the combined analyses or accepting that Kenya contributes with more facilities (and more health providers) compared to Tanzania? Would you be able to give me advise?
Re: Need to calculate country weight based on facilities when combining two SPA surveys? [message #10474 is a reply to message #10397] Wed, 27 July 2016 09:58 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member
Following is a response from DHS Stata Specialist, Shireen Assaf:

I would avoid pooling the data from the two surveys you mention, and run the model for each country/survey separately. If you run the models in this way, there is no need then to calculate country weights.
Re: Need to calculate country weight based on facilities when combining two SPA surveys? [message #10505 is a reply to message #10474] Mon, 01 August 2016 01:43 Go to previous messageGo to next message
masjas
Messages: 4
Registered: July 2016
Location: Amsterdam
Member
Thanks for you answer - however I am interested in pooling data to enlarge the sample size for these explorative analyses with taking into consideration the correct weighting for country. Therefore my original question on how to do this. Thank you.
Re: Need to calculate country weight based on facilities when combining two SPA surveys? [message #10513 is a reply to message #10505] Mon, 01 August 2016 09:38 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member
Another response from DHS Stata Specialist, Shireen Assaf:

It is not possible to have random effects for country if you only have two countries. Thus my suggestion to perform the analysis separately for each country. The same sizes are large enough for your analysis. If you would like to have random effects for facility and client then please see the post : weighting for Services Provision Assessment data by gizachew.

Previous Topic: Weighting variable in Pakistan 2012/13
Next Topic: which weight to use
Goto Forum:
  


Current Time: Thu Mar 28 18:57:49 Coordinated Universal Time 2024