| Re: Missing values in 2014 Bangladesh DHS [message #10120 is a reply to message #10119] |
Wed, 29 June 2016 10:51  |
Bridgette-DHS
Messages: 3043
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
This survey has complete birth histories for all children ever born to the women in the sample. All the births are included in the IR file, attached to the women, and as separate records in the BR file. The KR file is limited to all births in the past five years. Those are the births for which the child health information is collected. Some of the health information is only obtained for the youngest or more recent birth in the past five years.
It will be much easier if you use the KR file rather than the IR file. Then you can enter, for example, "tab m14 bidx,m" and you will see immediately that the number of ANC visits is only obtained for bidx=1, i.e. the most recent birth in the past five years. It is not obtained for any other births. You can confirm this by looking at the questionnaire.
The code "." in DHS files does not mean "missing". It means "not applicable". That is, the question was not asked about the children who have that code.
There was apparently some subsampling in this survey. Some households were skipped for these questions. You will see the same thing with m15, place of delivery. About 1/3 of the households were skipped for these items. Subsampling is always done randomly so it will not introduce any bias.
The documentation of this kind of subsampling is always provided in chapter 1 of the main report. Please look for it. If you cannot find it described in chapter 1, please let me know.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
