Generate new variable

[Lina Plotnikowa]

Hey,

in the panel study I am working with, there are two times the same variable (t34005a & t34005b). The first is only asked to people who no longer live at home and the second is only asked to those who still live at home.
I would like to create a new variable from this so that I have all the cases.

My idea was the following:
// restrict variable 2 to the cases who live either with the father or with the mother.


keep t34005a if t743024==1 | t743025==1

then the same for the variable of persons who no longer live at home:

keep t34005b if t743024 <>1 | t743025<>1

and then merge, maybe with append?

I'm a complete newbie and this is just a rough sketch.

Thanks a lot for the answers!

[Nick Cox]

You can keep a variable or you can keep observations but you can't mix syntaxes. Stata doesn't support <> as an operator; perhaps you mean != or ~=.

But keep sounds wrong any way. keep can be very destructive.

Also merge and append are for entire datasets, not variables.

If I understand you correctly this is closer to what you need in which you take some values from one variable and the other values from another variable

gen wanted = cond(t743024==1 | t743025==1, t34005a, t34005b)

That is equivalent to

gen wanted = t34005a if t743024 == 1 | t743025 == 1
replace wanted = t34005b if missing(wanted)

Naturally you should choose a variable name closer to what this represents.

gen wanted2 = max(t34005a, t34005b)

might even work.

If that doesn't help, please visit https://www.statalist.org/forums/help#stata and show us the results of

dataex t743024 t743025 t34005a t34005b

[Lina Plotnikowa]

Thank you very much! It worked

Have a nice day!