Truncating numerical dates to years

[Oliver Clarke]

I am trying to merge a few datasets, and one of them uses very specific date information in a format that is not, as far as I understand, legible to STATA as a date.

Specifically, the observations look like this:

start end
19460101 19461129
19460102 19460415
19460119 19460509
19460305 19460331

It is YYYYMMDD, such that 19460101 translates to January 1st, 1946.

The trouble with this is that the other data sets do not use this specific format, and indeed, I'd rather just use year.

Is there a way to generate new variables "start_year" and "end_year" such that they only include the first 4 digits of "start" and "end" respectively?

[Nick Cox]

Are those variables numeric or string?

[Nick Cox]

The question in #2 wasn't answered but remains pertinent. If for example start is string then a guess is

gen year = real(substr(start, 1, 4))

and if it is numeric a guess is

gen year = floor(start/10000)

[Nick Cox]

See also https://www.statalist.org/forums/for...year-from-date