Handling non-time varying variables in a longitudinal data

[PhanS]

Hello, Good day everyone !

I appreciate your help the below situation. Thanks you in advance. Phan S. 

data demo (drop=x);
retain id; do i= 1 to 3;
input id int age sex race gh mh;
if not missing(x) then id=x;
output;
end;
cards;
1 1 45 1 2 56 48
1 2 46 1 2 57 48
1 3 47 1 2 67 60
2 1 32 0 1 67 55
2 2 . . . 67 55
2 3 . . . . .
3 1 55 1 3 64 50
3 2 . . . 60 47
3 3 . . . 58 46
4 1 38 0 1 50 44
4 2 . . . . .
4 3 . . . 54 46
;
run;

*interview (int) every year, id-1 was recorded age, sex, race, gh and mh data but not other id did so.
1) I want to calculate age, sex and race for id-2, 3 and 4
2) id-2, and id-3 did not have gh and mh data in int3 and int2, respectively, how I handle this situation for age, sex and race

[Kurt_Bremser · Accepted solution]

One method would be to keep the values in retained variables, another is this:

data demo;
input id int age sex race gh mh;
cards;
1 1 45 1 2 56 48
1 2 46 1 2 57 48
1 3 47 1 2 67 60
2 1 32 0 1 67 55
2 2 . . . 67 55
2 3 . . . . .
3 1 55 1 3 64 50
3 2 . . . 60 47
3 3 . . . 58 46
4 1 38 0 1 50 44
4 2 . . . . .
4 3 . . . 54 46
;
run;

data help;
set demo;
by id;
keep id age sex race;
if first.id;
run;

data want;
merge help demo(drop=sex age race);
by id;
run;

[PhanS]

Dear Kurt,

Thanks for your help. 

After running, data merged well.

However, I notice that age for ID-1 changed to 45 in 3 interviews cycle. Because each interview was conducted every year, age of participants should be gained one-year age when interviewing.

I am happy to hear more from you or anyone. 

Appreciated.

Phan S.

[Kurt_Bremser]

Then you should handle age separately, by incrementing from the first record.

[PhanS]

Dear Kurt,

Thank you.

Phan S.