The DHS Program User Forum: Merging data files

Home » Data » Merging data files » Merge BR to PR (Merge BR to PR files for all survey rounds in Africa)

Show: Today's Messages :: Show Polls :: Message Navigator

Merge BR to PR [message #16733]

Mon, 25 February 2019 12:10

NZurlinden
Messages: 2
Registered: February 2019

Member

I work with all survey rounds in Africa. I want to merge the BR files to the PR files, but many children from BR are not merged to PR. I have looked at all posts in this forum related to merges, but have not found any solution to my problem.
From BR, I use caseid (and change it to hhid by stripping the last two characters), v003 (respondent's line number) and b16 (child's line number). From PR, I use hhid, hv003 (respondent's line number), and hvidx (child's line number). (Additionally, I have a unique identifier for the survey round). I exclude all cases with b16==. or b16==0 from the BR files.
I merge the BR to the PR files using the unique identifiers mentioned above.
I have around 2 mio children in my BR files. However, only around 1 mio are merged to the PR files.
I merged the two files only using hhid (dropping any duplicates), and most observations were merged hence, the problem is most likely not hhid.
I need to merge the BR to the PR files, because there are some variables that are not in the PR files, such as whether the child is a twin (b0), and the preceding birth interval (b11).
Does anyone have experience with merging BR to PR files, some possible solutions, or explanations why the merge does not work for so many children? Any hint would be much appreciated!
Many thanks!

Report message to a moderator

Re: Merge BR to PR [message #16837 is a reply to message #16733]

Thu, 07 March 2019 12:56

Bridgette-DHS
Messages: 3214
Registered: February 2013

Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

You want to match hv001 hv002 hvidx in the PR file with v001 v002 b16 in the BR file. No need to involve hhid or v003. Things to look out for: hv000/v000 is not a unique identifier of a survey; and some surveys in some countries in Francophone Africa have sub-household identifiers.

The BR file includes all the children ever born. Some of them can be 30+ years old. Children with b16=0 or b16=. have died or are not in the same household as the mother. There are MANY such children if you use the BR file.

As a secondary matter I recommend reducing the PR file to the variables you need. Otherwise you will be carrying along a lot of duplicated information.

Report message to a moderator

Re: Merge BR to PR [message #17719 is a reply to message #16837]

Mon, 13 May 2019 12:56

geoK
Messages: 56
Registered: May 2014

Senior Member

Hello, I am having a similar issue here and struggling to understand how to merge parents' information to child information, when the child is not in the BR file, but is in the PR file. For example, I had to use PR file as base for calculating all children of school age who are attending school, and merged it with BR. Now, I'd like to attach to those children all information about their parents, as well as households. How can i associate these info about mother, father, etc.. to children who are only in PR? Hope this is clear. Thanks.

Report message to a moderator

Previous Topic:	Merging Malawi MIS 2017 Datasets
Next Topic:	How to merge DLHS 4-Village data and DLHS 4-EW data

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Dec 19 07:09:59 Coordinated Universal Time 2024