| Re: Comprehensive HIV Knowledge [message #24541 is a reply to message #24535] |
Fri, 27 May 2022 13:44   |
Janet-DHS
Messages: 687
Registered: April 2022
|
Senior Member |
|
|
Following is response from DHS Research & Data Analysis Director, Tom Pullum:
I will describe the construction of this variable in terms of the women's (IR) file for this survey and Stata. First you have to find the relevant variables. Here they are, in terms of the "describe" command and what it produces:
. describe v754cp v754dp v756 v754jp v754wp
storage display value
variable name type format label variable label
------------------------------------------------------------ ------------------------------------------------------------ ---
v754cp byte %8.0g V754CP reduce risk of getting hiv: always use condoms during sex
v754dp byte %8.0g V754DP reduce risk of getting hiv: have 1 sex partner only, who has no other partners
v756 byte %8.0g V756 a healthy looking person can have hiv
v754jp byte %8.0g V754JP can get hiv from mosquito bites
v754wp byte %8.0g V754WP can get hiv by sharing food with person who has aids
You could find the exact wording of the questions in the questionnaire, which is included in an appendix. Then you have to find the codes for these five variables. They are all coded the same way, like this:
. label list V754CP
V754CP:
0 no
1 yes
8 don't know
Then, to match table 12.3.1 in the report, the variable is constructed with these two lines:
gen comp_know=0
replace comp_know=1 if v754cp==1 & v754dp==1 & v756==1 & v754jp==0 & v754wp==0
To confirm that this matches the table, do this:
. tab comp_know [iweight=v005/1000000]
comp_know | Freq. Percent Cum.
------------+-----------------------------------
0 | 4,069.3388 61.46 61.46
1 | 2,551.6612 38.54 100.00
------------+-----------------------------------
Total | 6,620.9999 100.00
This gives the number of cases (6621) and the relevant percentage in the bottom row of the table (38.5%).
I believe you were thinking that you would add up 1's and 0's and get a total, and then find a cutoff for that table. This variable has a more complicated definition. You have to specify that the first three components are exactly 1 and the last two are exactly 0. You can't construct it just with addition.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
