A common rule of thumb in social science is to have at least $30$ observations for regression (without any complicated effects). Does someone know such rule of thumb for GAMs? Are $50$ observations enough for a simple additive model with two parametric terms and two smoothers?
Asked
Active
Viewed 709 times
5
-
See discussion at http://stats.stackexchange.com/questions/56194/is-it-possible-to-use-a-two-sample-t-test-here/56211#56211 – Dale M Apr 16 '13 at 23:24