Hi all,

I am working with balanced panel data covering the period of 32 consequent years. I am analyzing the effect of marital status on the life satisfaction. I am treating the variables as cardinal since reg and ologit provide more or less the same results, also Hausman points towards FE rather than RE.

However, I am not sure if I need to include year effects (i.year) in my regression. Currently I am running the conditional fixed effect model (xtlogit) with year dummies and it takes significant time and seems to never provide results. I didn't have this problem before adding the year dummies and I wonder now if that might be the trigger or the big massive of data.

In general answering the question of life satisfaction would require or not controlling for the year effects ?

Thank you!

Kind regards,
Gabriela