in long data, replace var1==_n with other variable by id

[LarsFolkestad]

Dear Listers

I would like to replace a variable by the first occurrence of another variable.

input id first second
1 50.1 50
1 50.1 51
1 50.1 52
2 60.3 60
2 60.3 61
2 60.3 62
end

i want second[_n] to be replaced by first by id

so i tryed

bysort id: replace second=first if second==second[_n]

that just changed all second values to : second==first

so i tryed

bysort id: replace second[_n]==first[_n]

that gave me a weights not allowed error.

Hope you understand what i want and can tell me how to do it.

Lars

[Nick Cox]

I am not clear on what you want, but

1. _n always means "the current observation" so second and second[_n] will always be identical, regardless of whether you are using by:

2. Subscripts can't appear before the = sign in a replace statement, so any square brackets will be thought of as an attempt to use weights, not allowed here either.

You may want to consider whether

by id: gen third = first[1]

helps you. Otherwise show a worked example of what your data will look like when done.

[LarsFolkestad]

I understand your confusion, to name the variables first and second was a bad idea.

I want to replace the first value of B with the value of A so that the data looks like this:

input id A B
1 50.1 50.1
1 50.1 51
1 50.1 52
2 60.3 60.3
2 60.3 61
2 60.3 62
end

I think i figured out how to do that on the bike ride back from work.

input id A B C
1 50.1 50 52.3
1 50.1 51 52.3
1 50.1 52 52.3
2 60.3 60 63.3
2 60.3 61 63.3
2 60.3 62 63.3
2 60.3 63 63.3
end

sort id B
bysort id: gen seq=_n
replace B=A if seq==1

and if i want to change the last value of B with C

sort id B
bysort id: gen last=_N
replace B=C if last==seq

…and they say fresh air is good for you.

Lars

[Nick Cox]

Note that you don't need the extra variables. Consider (e.g.)

bysort id (B) : gen first = A[1]