Petra and Sarsmi are correct, as near as testing can indicate. Somewhere, what the devs think is the normal state for ore sites, and what actually occurs, went their separate ways before AoS - if it ever was true.
I took 3 sets of ore books before the ore randomization, condensed them down into 1-2 books per ore of spots that give 60% or more colored ore as a human, and while the ore types now change, the colored to iron ratio did not (well, you got to use a garg pick or Prospector tool to get colored ore if it goes iron, but it will still be the expected ratio recorded before, not a flat 50%).