The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Other countries » Rwanda DHS 2010: Discrepancy in result of Hb measurements
Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #725] Sun, 01 September 2013 10:49 Go to next message
richardn is currently offline  richardn
Messages: 4
Registered: August 2013
Location: Berlin
Member
Dear DHS Team,

I'm using the RDHS 2010 datasets to look, among other things, at malaria and hemoglobin in children between 6-59 months old. I'm working with STATA.

According to the children's recode file (rwkr61fl), 3733 children had their Hb measured (variable hw55).

Because I need malaria results only available in the individuals' recode (rwpr61fl), I merged the 2 datasets.

The PR recode comes with a corresponding variable for the outcome of Hb measurement (hc55). Now, according to this variable, 4068 children had their Hb measured. That's 335 more children than in the KR file.

I looked at all possible reasons I could think of and have not found a plausible explanation for this discrepancy. Can anyone help?

Thanks!!
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #763 is a reply to message #725] Thu, 12 September 2013 03:42 Go to previous messageGo to next message
alena is currently offline  alena
Messages: 1
Registered: September 2013
Member
Hello Richard and DHS team,
I'm facing similar problem for 2006 Niger DHS data. I'm also working with the KR dataset, permorming an analysis on the subgroup of children between 6 and 59 months with available Hb measurement. In KR dataset they are 3556 children 6-59 months old, 3206 of them with available Hb measurement. In the Niger DHS report the analysis of the child anaemia is done on 3732 children 6-59 months old. I cannot find the reason for such a difference.
Would be great if somebody could help on this issue.

Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #776 is a reply to message #725] Mon, 16 September 2013 03:40 Go to previous messageGo to next message
HanR is currently offline  HanR
Messages: 19
Registered: March 2013
Location: Netherlands
Member
Hello Richard and Alena,

Thanks for using DHS data. A few things to consider here. In the DHS Final Reports anemia is analyzed (in general) using all eligible children at the household level. A children's recode file is based upon children from de facto eligible women who were successfully interviewed. If you want to replicate the anemia figures from the DHS Final Reports you will have to use the household recode data. Please note and consider the following:
HW55 -> is presented at the individual level, ie children of defacto female respondents.
HV042 -> at the household level determines whether a household was included in an anemia (sub)sample.
HV120 -> from the household schedule indicates whether a child was eligible for hemoglobin measurement.
HC53 -> gives for children at the household level the hemoglobin measurement.
HC56 -> gives the same, but then adjusted for altitude.
HC57 -> gives the anemia level.

The difference of children measured for anemia at the household and indiviual levels may then come from:
- mother of the child does not live in the household
- mother of child does live in the household but was not interviewed
- mother of child was interviewed but was not a defacto member of the household
- mother of child does live in the household, was a defacto member, but not eligible for interviewing (outside the 15-49 years of age range).

Good luck with your analyses.
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #778 is a reply to message #776] Mon, 16 September 2013 11:32 Go to previous messageGo to next message
richardn is currently offline  richardn
Messages: 4
Registered: August 2013
Location: Berlin
Member
Dear Han,

Thanks a lot for this explanation. That could indeed explain the discrepancies that I have seen. If I understand you correctly, the PR file (which is based on household level data) would also reproduce the anemia analyses, is that right? My results with the PR recode have been consistent with the published report.

As a general comment, what is the best place too look for that kind of informaition? I have the recode manual, the recode map and the questionnaires used but still I had not been able to find the origin of the differences.

Regards,
Richard
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #786 is a reply to message #778] Fri, 20 September 2013 11:23 Go to previous messageGo to next message
HanR is currently offline  HanR
Messages: 19
Registered: March 2013
Location: Netherlands
Member
Dear Richard,

Indeed the PR file (household members) is in fact also a linked data set where the unit of analysis (the case) is one person in the household schedule and has, if applicable, height and weight and anemia data attached to it. A good place to look if you encounter these complexities is the questionnaire and also very useable is the DHS Guide to Statistics, downloadable from the Measure site at: http://www.measuredhs.com/publications/publication-dhsg1-dhs -questionnaires-and-manuals.cfm

Regards,

Han
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #787 is a reply to message #786] Fri, 20 September 2013 12:24 Go to previous messageGo to next message
richardn is currently offline  richardn
Messages: 4
Registered: August 2013
Location: Berlin
Member
Thanks Han!
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #2543 is a reply to message #725] Thu, 10 July 2014 04:29 Go to previous messageGo to next message
anneclaireclaire is currently offline  anneclaireclaire
Messages: 11
Registered: May 2014
Location: Italy
Member
Dear all,
I am trying to replicate the statistics on anemia presented in the Final 2010 DHS report for Rwanda. I have slight discrepancies with what is in the report (p154). It apparently comes from the fact that the DHS analysts have excluded some data. I have more observations than what is presented in the final.
Where can I have the information on what the cleaning procedure was ? which observation have been excluded.

Many thanks in advance
anneclaire
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #2544 is a reply to message #2543] Thu, 10 July 2014 05:35 Go to previous messageGo to next message
HanR is currently offline  HanR
Messages: 19
Registered: March 2013
Location: Netherlands
Member
Dear Anneclaire,

- is your analysis at the household level?
- do you consider children 6-59 months of age
- are your children in the analysis "defacto" (ie stayed at the household the night before the interview)

In general we do not exclude data or do a whole lot of "cleaning". We are particular about our population base though, which in this case is all defacto children living in selected households, 6-59 months of age.

Good luck with your analysis.

Han.
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #2551 is a reply to message #2544] Fri, 11 July 2014 03:46 Go to previous messageGo to next message
anneclaireclaire is currently offline  anneclaireclaire
Messages: 11
Registered: May 2014
Location: Italy
Member
dear Han ,
thank you for your quick answer!
-yes, i used household level (the PR recode file)
- for the two other , i now excluded less than 6 months children and not " de facto" children,i am closer

But i still have a very slight difference . For male children observation, i have 2026 children whereas you have 2037, i have 1983 female children and you have 1999.

I use this inclusion condition (Stata language)when calculating the descriptive statistics
hc57!=9 & hv103==1 & hv103!=9 & hc1>=6

Anything else to suggest, or should i consider taht it is enough close to use the data without further cleaning?
thanks,
ac
Re: Rwanda DHS 2010: Discrepancy in result of Hb measurements [message #2552 is a reply to message #2551] Fri, 11 July 2014 04:24 Go to previous message
HanR is currently offline  HanR
Messages: 19
Registered: March 2013
Location: Netherlands
Member
Dear Anneclaire,

You should be able to exactly replicate the report figures. Please consider the following filters:
- HV103 = 1 (defacto)
- HV042 = 1 (household selected for anemia testing)
- HC1 in 6:59 (age of children)
- HC55 = 0 (result of measurement)
- HV005/1000 000 - de correct weighting for your observation in this case (household weight divided by 1 million).

Good luck!

Han.
Previous Topic: Ghana 2003 - Height and Weight Scores Data
Next Topic: Rwanda PR recode h023 h024 districtt and region variables
Goto Forum:
  


Current Time: Fri Mar 29 04:32:20 Coordinated Universal Time 2024