Understanding Society User Support

Dear colleagues,

I intend to generate a variable that captures the age at which individuals enter into parenthood. For this purpose, I'm using two derived variables in UKHLS:

ch1by_dv: date of birth of eldest child (year)
doby_dv: date of birth of the interview (year)

With these two variables, I proceed as follows for deriving the age of entry into parenthood

gen age_first_child= ch1by_dv2-doby_dv2

Before, I ensure that some marginal categories of the original variables (-9, -20...) are disregarded. I create new variables, similar to the UKHLS-derived variables, for this purpose:

recode ch1by_dv -9=. 0=., gen(ch1by_dv2)
recode doby_dv -9=. -20=., gen(doby_dv2)

At the end of the process, I found out that the new variable ('age_first_child') has some (very few) negative values. I explore these cases and I find out that this is because the year of
having the first child (ch1by_dv) is BEFORE the year they were BORN (doby_dv). Since they are derived variables, I am rather puzzled.

Could you, please, help me understand these cases and what to do with them?

Thanks for your attention

And kind regards

Luis Ortiz