将虚拟变量与R回归

Question

I'm wondering how I can do a fixed regression while having some data points correspond to the aggregate dummy category "nonindustrialized" and the others correspond to their individual country names. I first ran a regression:

reg1 <- lm(birthrate ~ country*year)

and would like to subset the "country" data into nonindustrialized but leave all the industrialized data points as disaggregated. I made a logical TRUE/FALSE column for industrialized, but can't figure out how to subset it correctly without just getting it into two aggregated groups. Is there a way to just do it for the FALSE points and to have all the other points as individual countries?

Thank you!

Answer 1

It's not entirely clear from your question, but I'm assuming your dataframe is in long form and looks something like this:

country<-(rep(c("A","B","C"),4))
birthplace<-rep(c("x","y"),6)
year<-c(2001:2012)
df<-data.frame(country,birthplace,year)

> df
   country birthplace year
1        A          x 2001
2        B          y 2002
3        C          x 2003
4        A          y 2004
...

In that case, you can easily add a new column that either defines each line as nonindustrialized or else gives the original country value:

df$country.agg<-ifelse(df$country=="A"|df$country=="B","nonindustrialized",as.character(df$country))

Now you can use this column in your regression, which will pool all nonindustrialized countries into one category. Is this what you're looking for?