The DHS Program User Forum: Kenya » ANC components - discrepancies with final report usingGitCode

Home » Countries » Kenya » ANC components - discrepancies with final report usingGitCode

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

ANC components - discrepancies with final report usingGitCode [message #29332]

Fri, 31 May 2024 04:27

cippi
Messages: 4
Registered: June 2018
Location: United Kingdom

Member

Dear Staff, I have been trying to match the report and Statcompiler figures for ANC components, but the outputs given when i use the GitHub code give half the % that what it is reported on the final report and Statcompiler.

Specifically, I am interested in the following variables: ANC urine sample taken (m42d_1), blood sample taken (m42e_1) and took iron (m45_1), and I am using the 2014 Kenya dataset.

When i do a simple tab of m42d_1, m42e_1 and m45_1 then i get the correct percentages (same as Statcompiler). The denominator (not weighted) is 6764, and I know these should be women who have received ANC in the previous 2 (or 5 years) and the question is asked for their last birth only.

When i run the code and construct variables rh_anc_urine, rh_anc_bloodsamp and rh_anc_iron my denominator is almost double, hence i get half the %. I am not sure what I am doing wrong here, and I'd appreciate help on identifying the correct denominator and how to select it, in order to get the correct % for ANC components.

Thank you!

Report message to a moderator

Re: ANC components - discrepancies with final report usingGitCode [message #29336 is a reply to message #29332]

Fri, 31 May 2024 08:49

Trevor-DHS
Messages: 803
Registered: January 2013

Senior Member

Kenya 2014 DHS used a split sample with a long questionnaire and a short questionnaire. Some indicators can only be calculated for the half of the sample that used the long questionnaire, and that is probably the case for this indicator. Thus, you need to select for cases that used the long questionnaire. You can use v044 to identify the long questionnaire, as follows:

gen longqre = 1 if v044 != .

then select those cases with longqre == 1

Report message to a moderator

Re: ANC components - discrepancies with final report usingGitCode [message #29338 is a reply to message #29336]

Fri, 31 May 2024 11:28

cippi
Messages: 4
Registered: June 2018
Location: United Kingdom

Member

All clear, thank you!

Report message to a moderator

Previous Topic:	De jure region of residence (v139) - Kenya 2014
Next Topic:	Mismatch When Merging KR and PR recode files due to disparities in age variables

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Tue Nov 26 07:01:09 Coordinated Universal Time 2024