They are approximately equally accurate methods...but only if you use all five of the characteristics I measured (though you could probably use either growth rate or weight instead of both; they returned pretty similar results).
I'm still writing and I distracted myself trying to do some math I hadn't done since college, but figured I'd drop this beauteous info on y'all to tide y'all over until I can talk to the person in this house who actually likes math about the problem. I only looked at Week 6-9, since it became pretty obvious early on that that was the sweet spot; before that the sexes were just too similar, and this is how the accuracy came out, with no answer (for the guessers 50/50 splits were no answers) being regarded as wrong for both methods:
Week 6:
Humans guessing - 63%; Crunching numbers - 58.33%
Week 7:
Humans guessing - 79%; Crunching numbers - 70.83%
Week 8:
Humans guessing - 87.5%; Crunching numbers - 87.5%
Week 9:
Humans guessing - 83.33%; Crunching numbers - 87.5%
The value of this info (to me, anyway) is that you can measure your birds independently...you wouldn't need to rely on finding people to peep at your birds and tell you what they see there.
The method with the highest accuracy (of course y'all know I tried it multiple ways) was to evaluate the birds Week 6, 7, 8, and 9 (though since the highest accuracy for both methods is reached at Week 8, there's probably little reason to include 9); birds that have any of the five measured characteristics 1 standard deviation below average are female. If they don't have any in that range, but have one or more that are 1 standard deviation above average, they are male. If you get a result of female during any of those weeks, ignore future or previous male results for that individual.
Y'all can consult my tables for my averages once I slap this Word document on the bottom but it'll probably be useful if someone or everyone tries to collect more data for this data pile so we can be as accurate as possible.