The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Merging data files » dhs data set
dhs data set [message #10991] Sun, 16 October 2016 12:10 Go to next message
phres110 is currently offline  phres110
Messages: 39
Registered: October 2014
Location: korea
Member
Hello
i am using DHS of 5 countries for my research i am new to dhs, i have the following questions:
1. i am using maternal health indicator( delivery assistance) and background characteristics and wealth index for investigating the relationship.Kindly guide me which data set i have to use that has all the desired variables. i am using birth data set for delivery assistance it has also some variables related to age and education but i am not sure ...should i use the others files too? Should i need to merge the files..if so which one?
2. i made some tables with the help of SAS compiler that gives me totally different results in SPSS and SAS WHY?
3. For wealth index of the mother of delivered babies which variable for wealth index i have to choose?
thank you



dhs110
Re: dhs data set [message #10994 is a reply to message #10991] Mon, 17 October 2016 09:29 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member

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


Quote:
If you want women to be your units of analysis, then you work with the IR file and just the most recent birth, which has index _1. For example, m3a_1 tells you whether a doctor was in attendance at the birth of the most recent child. When women are the units, you have to pick one birth.

If you want children to be your units of analysis, then you use the KR file. For example, open a KR file and enter "tab bidx m3a". Here bidx is the order of the birth in the past five years (bidx is 1 for the youngest child) and m3a tells you whether a doctor is in attendance at the birth of that child. In both the IR and KR files, the wealth index (in quintiles) is v190.

For some surveys the information about the delivery is only given for the most recent birth in the past five years, not for all births in the past five years.

I can help you with Stata but not with SPSS or SAS.


Re: dhs data set [message #10996 is a reply to message #10994] Mon, 17 October 2016 10:20 Go to previous messageGo to next message
phres110 is currently offline  phres110
Messages: 39
Registered: October 2014
Location: korea
Member
Thank you very much...............so nice of you

dhs110
Re: dhs data set [message #10997 is a reply to message #10994] Mon, 17 October 2016 10:24 Go to previous messageGo to next message
phres110 is currently offline  phres110
Messages: 39
Registered: October 2014
Location: korea
Member
Dear
i want to use cesarean section as my outcome variable and that is M 17 i want to know if i use BR files would it be enough for all variables? OR should i merge any file?
i am confused as i saw a lot of same variables in the same files like v102 and v 025 both are for type of place of residence, similarly v101 and v024 both for region....which one i have to use?
my last question is why results are so different from sas spss and stat compiler using the same data files?
Thank you


dhs110
Re: dhs data set [message #11000 is a reply to message #10997] Mon, 17 October 2016 14:08 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Quote:
Your outcome is in both the KR and BR files, but I recommend using the KR file, not the BR. Virtually all of the IR variables are included in the BR and KR files. You should not need to do any merges.

The results from Stata, SPSS, and SAS should agree when you are using the same procedure. Please inform me of any specific example where they do not agree.

Re: dhs data set [message #11005 is a reply to message #11000] Tue, 18 October 2016 01:24 Go to previous messageGo to next message
phres110 is currently offline  phres110
Messages: 39
Registered: October 2014
Location: korea
Member
Dear DHS Experts
Thank you very much for quick and gentle replies, i have some more questions as i am new in DHS so i am sorry to bother you again and again but your answers are very valuable to me as they are making my research easy thanks again.i have these questions:
1.In PDHS 1990-91 BR files the wealth index variable V-190 is not listed from which file i can use that from household?
2. How to merge the file in this case how do i know i am using the wealth index of same women whose other variables i am using?
3. If i use IR files there are a lot of choices in my desired variable C/S like M -17-1, 2, 3,..... I think they are for all births, i am working on last birth so which one is showing the last birth?
SPSS ans SAS results are same, but both of them are different from stat compiler....why stat compiler show different results?
Regards and again thank you very much.


dhs110
Re: dhs data set [message #11014 is a reply to message #10994] Tue, 18 October 2016 12:19 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Quote:
The wealth index for the Pakistan 1990-91 survey is available in another file. The Stata version is PKWI21DT.ZIP. Please login to your account to download that file (or the SPSS: PKWI21SV.ZIP or SAS: PKWI21SD.ZIP). I can help with the merge commands in Stata, but not in SPSS or SAS, except to say that whhid in this file matches with hhid in the HR and PR files, wlthind5 is the usual hv270 or v190, and wlthindf is the usual hv271 or v191. hv270 and hv271 in the household file are the same as v190 and v191 in the IR/BR/KR files, and the same as mv190 and mv191 in the MR file.

Everyone in the same household has the same value of the wealth index. It's defined at the level of the household, using household assets. That is, it is determined by hhid or the combination of hv001 and hv002 in the household data or v001 and v002 for the woman. Beyond saying that, I don't know how to answer your question about knowing whether you have the right value of the wealth index.

The last birth is the one with bidx=1.

If you are trying to match with Stat Compiler, you must tell me exactly what number you are trying to match, and what you get instead. However, I can only help with Stata. If you are trying to match with SPSS or SAS, someone else will have to help.

Re: dhs data set [message #11022 is a reply to message #11014] Thu, 20 October 2016 00:48 Go to previous messageGo to next message
phres110 is currently offline  phres110
Messages: 39
Registered: October 2014
Location: korea
Member
Dear Sir,
Thank you very much for reply, and i am sorry again to ask so many questions.I have only query now....as far as i understand after reading DHS docs the BR file is the part of IR file having a separate record of birth histories if so, the results should be the same after running the same variables in both files. For example:
V024(Region) VS v208 (Births in last 5 yrs) should be the same in IR files and BR files but its not i tried many times but the data in both files gives different results...why?
Thank you in advance and Best Regards.


dhs110
Re: dhs data set [message #11048 is a reply to message #11022] Thu, 20 October 2016 16:16 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3016
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Quote:
The IR file has one record per woman. A variable such as v208 is defined for the woman, and this is the file you should use for that variable.

The BR file has one record per child, whether alive or in the household.

In the BR file, the data from the mother is attached to the child's record. That means the woman's data is repeated for as many children as she has had. Women who have had no children will be missing entirely. Therefore calculating the mean value of v208 in the BR or KR file is incorrect.
Previous Topic: One to one merging
Next Topic: Combining Household and Children recodes into one data file in STATA
Goto Forum:
  


Current Time: Thu Mar 28 07:41:38 Coordinated Universal Time 2024