Two way or one way fixed effect ??

[victor cid]

Hi,
test to choose between model two way fixed effec or fixed effect ordinary ? i use STATA 14

[Andrew Musau]

You should check for the joint significance of the time dummies. Use testparm.

webuse grunfeld
xtreg invest mvalue kstock i.time, fe
testparm i.time

A significant test statistic means that you reject poolability across time, and therefore you should favor two-way fixed effects over one-way fixed effects.