| Re: Anthropometric variables [message #22832 is a reply to message #22816] |
Mon, 17 May 2021 09:13  |
Bridgette-DHS
Messages: 3218
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:
In the first place where you have "wa3=0, there is a typo. You should have "wh3=0". However, there are several other things I would modify. You seem to be using "keep" in situations where it would be better to use the NA code, rather than dropping cases. You have to be very careful about the impact of dropping cases on the remainder of the program. Avoid inventing variable names such as "wa3". Here's how I would get "stunted", for example:
gen stunted=.
replace stunted=0 if hc70<9990
replace stunted=1 if hc70<-200
This is ok because hc70-hc72 have already been forced to be in the ranges -600 to +600, etc. Otherwise your recode would have to use the legal limits for the Z scores.
Perhaps the most important problem is that "overweight" would be correctly indicated by high values of hc72, rather than hc71. This may seem counter-intuitive, but that's how it's defined.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
