Why not just test it? Use a class/race combo with no stat mods, and "Reroll" a couple dozen times. You should be able to make a pretty good guess. Here's a quick start from me:
From 19 sets of statistics rolls:
Average: 11.719
Standard deviation: 3.058
Min: 6
Max: 18
I'm not certain, but that looks to me suspiciously like the good old "Roll 4d6 and drop the lowest" method of classic D&D.;
From the same 19 rolls, for resistances:
Average: 28.833
Standard deviation:4.955
Min:20
Max:40
I also note that the resistance rolls seem to come in groups - the greatest difference I've seen within any set of resistances is 6.
Lacking any official response from Dee, this would seem like a pretty good way to start.
------------------
I love deadlines. I like the whooshing sound they make as they fly by. - Douglas Adams
Only member of the webboard to gain 2 karma points by writing a flame, and openly atychiphobic since 2000.