Re: Trends in vaccination coverage and factors for complete vaccination [message #19697 is a reply to message #19678] |
Tue, 04 August 2020 09:51 |
Bridgette-DHS
Messages: 3042 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
For the first question, I recommend combining the two surveys into one file with "append", constructing a variable "survey" to identify which was which. The first survey could have survey=1 and the second survey would have survey=2. Then you would basically do exactly what you would do with just one survey, to see whether there is a difference between urban (v025=1) and rural (v025=2) areas. In Stata you would need to make the svyset and svy adjustments, which are described in other postings.
For the second and third questions, you would first look at the main reports on these surveys to see what covariates were used, and at publications using DHS data and other data. You are limited to variables in the data, so the list of candidate covariates (potential risk factors) is not very long. They include region, urban/rural residence, education of the mother, wealth quintile, etc. Cluster-level spatial covariates could be helpful. The process would not be very different from what you would do with any outcome.
|
|
|