The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Merging data files » coverage of variables v717 and mv717 in Mali V and VI waves
coverage of variables v717 and mv717 in Mali V and VI waves [message #22154] Mon, 08 February 2021 10:25 Go to next message
bcleite is currently offline  bcleite
Messages: 7
Registered: June 2018
Member
Hi,

I am using data of the waves V and VI of the Malian DHS survey. I am aggregating the data from the couple (respondent)'s recode containing the variables v717 (mv717) at the household (v007-v001-v002) level to then merge it to the children recode at the household level. I noticed that, on average, about 75% of the households within each cluster does not have data on occupation (missing v717 and/or mv717).

Is there any explanation for so much missing data?

Thanks and best,
Bruno
Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22174 is a reply to message #22154] Tue, 09 February 2021 08:56 Go to previous messageGo to next message
Bridgette-DHS is currently online  Bridgette-DHS
Messages: 2166
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

When I look at MLCR53FL.dta, fewer than 1% of the couples are missing on v717 or mv717.

If you merge the parents with the children, the children will not have codes v717 or mv717 and they will be assigned a dot (meaning NA rather than missing). Could it be that you are just seeing those cases?

Users are asked to identify files by the year of the survey or (preferably) filename.
Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22177 is a reply to message #22174] Tue, 09 February 2021 09:38 Go to previous messageGo to next message
bcleite is currently offline  bcleite
Messages: 7
Registered: June 2018
Member
Dear,

thanks for your feedback. I illustrate my issue further (with the file names).

My goal is to match household-level information on work occupations from MLCR53FL.dta to the children's recode (MLKR53FL.DTA). I find that about 75% of the households (identified with v000 v001 v002 v007) in the children recode are not present in the couple recode.

An illustration of the code I am using (in Stata):

use MLCR53FL.DTA // load couple recode
gen agric = mv717==10 // dummy for male member working on agriculture
collapse (mean) agric, by(v000 v001 v002 v007) // average male individuals at the household level working in agriculture
save "temp.dta", replace

* merging to children's recode:
use "data/DHS MAL/data occupation/MLKR53DT/MLKR53FL.DTA", clear // load children recode
merge m:1 v000 v001 v002 v007 using "temp.dta" // merging it

tab _merge // about 75% of the households of the children are not present in the couple's recode

Does that expose my issue clearer?

Thanks in advance.

Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22181 is a reply to message #22177] Tue, 09 February 2021 10:47 Go to previous messageGo to next message
Bridgette-DHS is currently online  Bridgette-DHS
Messages: 2166
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

Is there some reason why you are using the CR file for this purpose? The construction of that file is very restrictive. Both the woman and the man have to identify each other as partners. Another issue is that in Mali polygamy is common. The men may only be matched with the first wife listed. These are possible reasons why so many children in the KR file do not merge with the CR file.

MLKR53 already contains v717, which is the stated occupation of the mother. All of the v variables in the KR and BR files refer to the child's mother. If you want to attach the father's data, you can do that if the father is in the MR file. You use the PR file to get hv113, the line number of the father if in the PR file, and then the man with that line number (changed to mv003) and the same cluster and household number as the child. It's a trickier merge than most but I provided the Stata code in another post.

If the father is not in the same household as the child, you could consider using the occupation of the household head (if male). It depends on whether you are thinking of the child in relation to the biological parents or to the adults in the household.
Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22192 is a reply to message #22181] Wed, 10 February 2021 02:40 Go to previous messageGo to next message
bcleite is currently offline  bcleite
Messages: 7
Registered: June 2018
Member
I see, using MLKR53 should be much easier then. If I want to use the occupation of the household head, which variable (and from which datafile) I should use? How would I link it to the MLKR53?

Thanks a lot.
Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22244 is a reply to message #22192] Wed, 17 February 2021 09:03 Go to previous messageGo to next message
Bridgette-DHS is currently online  Bridgette-DHS
Messages: 2166
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

You can merge information from the household head with the children in the household with the following steps. Open the PR file and enter "keep if hv101==1" . That will reduce the file to just the household head, one and only per household. Then "rename hv001 v001", "rename hv002 v002", "sort v001 v002" , and then "merge v001 v002 using KR.dta", where KR.dta has been sorted on v001 and v002. You don't need to specify the child's ID code (which is bidx) in this merge. Just using the cluster and household ID codes for the household head and the child (or children) in the household will do it. I recommend that you reduce to the variables you plan to use, before doing the merge. No need to carry along a lot of variables you don't plan to use.
Re: coverage of variables v717 and mv717 in Mali V and VI waves [message #22250 is a reply to message #22244] Thu, 18 February 2021 03:41 Go to previous message
bcleite is currently offline  bcleite
Messages: 7
Registered: June 2018
Member
that makes a lot of sense, thanks so much!
Previous Topic: MERGING HIV DATASETS AND WOMEN DATASETS, DHS 2016-2017, HAITI.
Next Topic: Merging variables of mothers' health status in IR file into KR file
Goto Forum:
  


Current Time: Tue Mar 2 15:40:26 Coordinated Universal Time 2021