The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Malaria » Missingness in strata variables (Madagascar MIS- 2011 and 2013)
Missingness in strata variables (Madagascar MIS- 2011 and 2013) [message #26720] Sat, 22 April 2023 16:28 Go to next message
cleonard is currently offline  cleonard
Messages: 2
Registered: November 2017

I am working with several DHS and MIS datasets and am referring to the DHS guide for statistics to calculate malaria prevalence in children under 5 ( m).

The issue I have come across is that for Madagascar 2011 and 2013 MIS the variable, hv022, is missing for each observation. I'm not sure how to incorporate the complex sample design into survey-weighted estimates when there is no value for the strata (hv022). The datasets I'm using are: MDPR61FL.dta and MDPR6AFL.dta

Also, child's age in months (hc1) is not present in the Rwanda 2017 MIS (RWPR7AFL.dta). I was surprised the variable was not found in the dataset, is there a way to calculate this?


Re: Missingness in strata variables (Madagascar MIS- 2011 and 2013) [message #26761 is a reply to message #26720] Wed, 26 April 2023 14:31 Go to previous message
Janet-DHS is currently offline  Janet-DHS
Messages: 393
Registered: April 2022
Senior Member
Following is a response from DHS staff member, Tom Pullum:

You are right--hv022 is missing in MDPR61 and MDPR6A. hv023, which is often equal to hv022, is also missing. I checked our master list of Strata and it says there is no stratum variable in any of the files for the 2011 and 2013 MIS's in Madagascar. In this case the default is combinations of hv024 and hv025. In Stata, enter "egen stratum_id=group(hv024 hv025)" and then use stratum_id in the svyset command.

I also see that hc1 is missing from the PR file for the Rwanda 2017 MIS (RWPR7A). In the KR file, b19 (age in months) is constructed using the date of interview and the date of birth. I have just confirmed that you can match b19 with "gen b19_test=int((mdy(v006,v016,v007)-mdy(b1,b17,b2))/(365.25/12 )) ". If you calculate b19_test with this command, however, you will find that there is one problem case--a child allegedly born on November 31, 2015. November only has 30 days, so mdy returns a dot for that case. You have to change the birthday for that child to November 30, 2015 to match b19 for that case.

In DHS surveys, the PR file includes the month, day, and year of the child's birth, and you could confirm that hc1 is constructed with the mdy() function as above. However, MIS's are not as complete as DHSs and RWPR7A does NOT include the month, day, and year of the child's birth. I don't see how you can construct hc1 to match b19. I do see that the file includes hml16a, "age in months (for children)". I hope this is equivalent to hc1, but I don't see a way to confirm that.
Previous Topic: Children Data NMIS 2015
Goto Forum:

Current Time: Thu Jun 8 00:51:29 Coordinated Universal Time 2023