About Multiple response variable

[Shamsudini Amidu]

Hi family,
I have multiples choice question that I splited them into 4 variable base on the command I came across, with a yes for option that were chosen and for the choice that were not chosen it indicates as missing response since that was not selected.

example,
saq4: Select the source of the information(Multiple response)
A. School(Selected)
B. Home.
C. Television(selected)
D. Other

So in column saq4= AB as text

After conversion,
using the command below that i came across;

foreach j in A B C D{
gen saq4`j'=regexm(saq4,"`j'")
recode saq4`j' (0=.)
la val saq4`j' yn
}

lab var saq4A "Does not visit home regularly"
lab var saq4B "Does not answer calls"
lab var saq4C "Does not attend social events"
lab var saq4D "No development"

I got four(4) additional newly created variable with response as seen below;

saq4A saq4B saq4C saq4D
yes . yes .

example of the structure after conversion, the rest,

OS. saq4A saq4B saq4C saq4D
1 . Yes . yes
2. yes . . .

.
.
.

20. . yes yes yes

I used the command
mrtab saq4A saq4B saq4C saq4D

Though I got some statistics but I was not too sure.

Because the examples that I came across, they used 0 for no and 1 as yes but in my case I have missing(.) as no and yes for the responses.
Can you please confirmed for me

I attached example data set for your perusal.
. dataex

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input int zabea_id str4 saq4 str6 saq5
 1 "D"   "ABCDE" 
 2 "D"   "ABCDE" 
 3 "D"   "ABCDE" 
 4 "BD"  "BDE"   
 5 "BCD" "ABCDE" 
 6 "CD"  "BDE"   
 7 "D"   "BDEF"  
 8 "D"   "C"     
 9 "D"   "ADEF"  
10 "D"   "DE"    
11 "D"   "DE"    
12 "A"   "DF"    
13 "D"   "DE"    
14 "D"   "DF"    
15 "D"   "DF"    
16 "D"   "ABCDE" 
17 "D"   "ABCDE" 
18 "D"   "ABCDE" 
19 "A"   "ABCDEF"
20 "D"   "ABCDE" 
15 "D"   "ABCDE" 
end

I would be glad if there is also another easy way to handle it I will appreciate that.

Thank you.

Sham.

[Nick Cox]

I am not quite sure what the question is here, as your data example contradicts much of what you say earlier. However,

foreach v in A B C D E F {
     gen saq5_`v'  = strpos(saq5, "`v'") > 0
}

gets you a bundle of (0, 1) indicator variables.

See https://journals.sagepub.com/doi/pdf...867X1101100308 for a round-up of functions that -- in the author's opinion -- every long-term user of Stata would benefit from knowing.

[Shamsudini Amidu]

Hi Nick,
Thank you very much Sir.

Sorry I for got to add this command

la def yn 1 "Yes"

before the focast loop.

I used only one variable saq4 to run the command.
However, I think I am okay. what you have provide makes it much easier. I will also make good use of the link you share to improve my understanding.

I am so grateful.