Following is another response from DHS Research & Data Analysis Director, Tom Pullum:
I see that only two codes were used for these variables: 1 "big problem" 2 "not a big problem". I suggest that you recode them as follows:
label define v467_rev 0 "not a big problem" 100 "a big problem"
gen v467b_rev=.
replace v467b_rev=0 if v467b==2
replace v467b_rev=100 if v467b==1
label values v467b_rev v467_rev
gen v467c_rev=.
replace v467c_rev=0 if v467c==2
replace v467c_rev=100 if v467c==1
label values v467c_rev v467_rev
gen v467d_rev=.
replace v467d_rev=0 if v467d==2
replace v467d_rev=100 if v467d==1
label values v467d_rev v467_rev
egen v467b_clusterpct=mean(v467b_rev),by(v001)
egen v467c_clusterpct=mean(v467c_rev),by(v001)
egen v467d_clusterpct=mean(v467d_rev),by(v001)
Then the three "clusterpct" variables will be the percentage of women within the cluster who say the respective issue is a "big problem".
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
