I want to perform a regression in a fixed-effects model. To construct such a model, I have multiple FLAGs, like the following:

y ~ x   z   FlagYear1   FlagYear2   FlagYear3   FlagCountry1   FlagCountry2

I want to perform another regression in which I have fixed effects for Year * Country, so that the model will be equal to this

y ~ x   z   FlagYear1Country1   FlagYear1Country2   FlagYear2Country1   FlagYear2Country2   FlagYear3Country1   FlagYear3Country2

As I have 26 countries and 8 Years in my model, so it would be very time-consuming to manually construct all the FLAGs. I know there is a command to perform this automatically in Stata, how can I do the same in R?

CodePudding user response：

If by 'FLAG' you are referring to 0/1 coded indicator variables (or dummy variables) then R has an easy way to enter all of these interactions into a forumla.

If you have factor variables country with 26 levels and year with 8 levels then you can use

y ~ x z country*year

and this will expand the factors into every combination of country and year.

Look at the documentation for formula to understand how this works.

If you already have the indicator variables then you could use

y ~ x z (FlagYear1 FlagYear2 FlagYear3) * (FlagCountry1 FlagCountry2)