I have another question about how to control for "fixed effect" more
efficiently.
For example, I have a dataset of the level of state, year (of survey)
and age. I would like to control for state specific effect, year
specific effect and age specific effect, but not interaction of them.
One way to do it is to put in state dummies, year dummies and age
dummies. However, sometimes when the dataset is large, such strategy
will make the dataset much much bigger and difficult to handle.
Moreover, their coefficients are not always of our interest. Then can
any expert here suggest a more clever way of doing such a job?
