Ln(dadprestige): is simply elasticity of x with y, so 1% change in x leads to a 0.15% change in y (although I found that this doesn't hold true if you say jump from a dadprestige of 10-20, i.e. a 100% change in dadprestige won't cause a 15% change in prestige).
The interaction term and nonwhite is where my trouble is:
A non-white will have -78.7% less prestige (-0.787*100)
a dad prestige of 50 will, on average, have a prestige rank which is -78.7% (-0.787*100)
and 0.2*x% more for a 1% change in dadprestige...
But even if I subtitute in lets say a dadprestige of 10, I find that my interpretations aren't actually true..