Dear stata users,

I'm a PhD student, trying to run a conditional logit estimation on stata with microeconomic data. I have macro data for conflicts per country per year, and micro data with surveys for around 10.000people per country per year (using Gallup world poll).

I tried several things to conduct this regression:

- clogit y x, group(o)
with o being a numerical code for origin countries. This estimation gives me the error 1400 such as "5,996 (group size) take 1,297 (# positives) combinations results in numeric overflow; computations cannot proceed r(1400);"

So I understood this was because there were multiple observations for each combinations of country/year. Knowing that I have a lot of observations per combinations, I also tried:

- xtset o year
xtlogit y x, fe(o)

The problem is when I try to xtset, (obviously) I have multiple observations per combinations of origin and year (survey), so I have an error message. And I cannot drop duplicates in terms of conflicts at the year/country level, since it would drop me the microeconomic data.

I tried to be as clear as possible, I hope it will be enough.
Do you know a way to tell stata I am using panel data at the country/year level, while still keeping multiple observations per country/year combinations?

Thank you in advance,
Best,
Killian Foubert