The DHS Program User Forum: Dataset use in Stata » Unique identifier in pooled DHS (countries and years)

Home » Data » Dataset use in Stata » Unique identifier in pooled DHS (countries and years)

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

Unique identifier in pooled DHS (countries and years) [message #971]

Tue, 10 December 2013 08:40

marion
Messages: 1
Registered: December 2013

Member

Dear all,

I pooled the birth history datasets for all phases and all countries. My total sample consists of roughly 6,000,000 observations. I created a unique identifier by combining v000+v001+v002+v003+bidx, as it is recommended in the DHS manuals. However this combination does not uniquely identify each line (=each child). There are more than 1,500,000 observations which are not uniquely identified by this combination of variable. My guess is that is problem could be connected to the presents of different waves within the phases (e.g. for Uganda phase 5 there are the datasets UGBR5HDT and UGBR52DT). Do I have to assign a coding for the different waves within the phases? I did not read anything about this requirement in the manuals so I am not sure how to proceed best. Do you have any advises?

Many thanks and warm regards,
Marion

Report message to a moderator

Re: Unique identifier in pooled DHS (countries and years) [message #975 is a reply to message #971]

Wed, 11 December 2013 14:41

Reduced-For(u)m
Messages: 292
Registered: March 2013

Senior Member

Hi Marion,

I don't have a great answer for how to use those variables to create a unique id number, but if all you want is a unique id (and don't want to use part of that ID to somehow group the obs by, say, country and round) then one easy way would just be:

gen id = _n

Then you'll get unique IDs of 1-N. If you have to match this to other data somehow, you could always sort it:

sort country wave something
gen id = _n

Or, if you want to check your intuition on what isn't unique, you could create a number of each "wave" within each "phase" and then add .01*wave (or 10000000*wave, however many 0s you need) to each identifier, and see if you have unique ID's.

Now, if there is some reason you need some more specific kind of ID number, then these might not work and you might need to wait for a DHS person to help.

Report message to a moderator

Previous Topic:	calculating median breastfeeding duration using current status data
Next Topic:	Pooled Datasets - Use of Svyset & regional controls

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Nov 21 16:43:47 Coordinated Universal Time 2024