I think there's also one other factor that hasn't been mentioned yet. You also have to take into account the network hash rate of the coin, and how lucky/well the pool your in is doing in that coin. From my own experience in the last few months of mining, this varies greatly coin to coin, and even pool to pool in the same coin. The only real way to compare algo's IMO would be for two coins to have the same network hash rate divided exactly the same over the same pools. The coins would also have to have the same block payout and block target times too, or it wouldn't be a accurate comparison.
No, you can just compare the BTC that you would get when mining and trading that on an exchange.