E. Marshall data(1 shot stop, torso hits only):
9mm Luger:
Federal 115 gr JHP +P+ 91%
.45 ACP :
Federal 230 gr. HS JHP, 96%
G Ellifritz data:
9mm One-shot-stop, any hit 34%
incapacitated by one shot (torso or head hit) 52%
.45ACP One-shot-stop, any hit 39%
incapacitated by one shot (torso or head hit) 51%
(and he added a caveat that over half of the 9mm data was FMJ ammo and that it is reasonable to assume that FMJ does not perform as well as expanding bullets)
Personally, I agree with Marshall that only the torso should be considered in a "one shot stop" evaluation.
This subject has been hotly debated for years because there are far too many variables to consider. There simply isn't any way to draw definitive conclusions without much more data... but clustered numbers are clustered numbers and they can be expected to allow comparisons between calibers within each study ... Provided that they're viewed as having a large margin of error. I would think AT LEAST 5%.
Intuitively, I like the big fat heavy slow bullet theory, but the data available doesn't really show that. Also, even if a 4 or 5 percent difference between 9mm and .45acp is true ... when you consider 16 shots versus 8 ... 5 percent or less difference in each shot starts looking really small.
The best conclusion I can draw is that as long as the bullet is carrying enough energy to form a significant wound channel, it'll do the job if the shot is placed correctly. As for me, I'm comfortable with the performance of .44 spl., 45ACP, 9mm, .40s&w, .357mag etc. .... 10mm, .357Sig, .44mag. are a little obnoxious in my book for people, and a little weak for bears.
I would suggest...Carry the biggest you can control easily and are comfortable with.