It's an interesting video but it's still got some personal bias when he's trying to make a point, and some odd "math".
For instance, at 1:45, he says "difference of 5 percent are nothing to get excited about, details too fine to catch in the study could easily swing things a few points in either direction" and the text says "5% or less is not significant". Later on, (7:21) when trying to prove that the .45acp is no better (and maybe worse) than other handgun rounds, he goes out of his way to point out that a 5% difference means it "lags a bit behind". He accents this point by saying that the .380acp actually BEATS the handgun average... but doesn't point out that it "beats" that average by only 6%... a mere 1% from what he called "not significant" earlier.
A few seconds later, around the 2:00 mark, he says that when it comes to stopping the attack "bullet size doesn't make any difference", but the bullet sizes he circles at the top of the chart only differ by .096", .380acp to .45acp and only constitute 3 calibers. That is simply not enough data to make such a sweeping claim.
It is interesting to note that the average (mean) number of shots needed to result in incapacitation is more than 1 but less than 2 for all gun types. Since number of shots is "digital", as in whole numbers, there is no difference between 1.2 and 1.8 shots... both are "2" (or more). The only way that statistic has any real meaning is if the Standard Deviation of various rounds is significantly different, and they don't give us that information.
He also says early on that all .22 data of all cartridge and firearm types is pooled into ".22LR" but then he uses that pooled data to "prove" that a .22LR rifle is massively less effective than a "real" rifle or shotgun. While this may in fact be true, it is not a valid conclusion from mixed data.
The biggest problem, though, is the entirely presumptuous extrapolation regarding .410 handguns starting around the 8:40 mark. First, he says "Ellifritz didn't gather data regarding .410 handgun shootings" but then goes on to make several points regarding what might have been, if he had.
First, you have to go back to where he said (in conjecture, no data supports it) that the reason a shotgun firing buckshot is more effective is because "each pellet carries the same energy as a typical handgun round", because.. well, first off that's just not true. 9 pellets with 1550 ft-lbs of energy total each carry about 171 ft-lbs. This is nowhere near a typical handgun round. However, even if we assume his statement to be true, the first correlation he makes is that the reason a .410 handgun might be half as effective as a shotgun is because it fires 5 pellets while the shotgun fires 9. Never mind that the TOTAL ENERGY of those 5 (at most around 300 ft-lbs) is less than TWO shotgun pellets and less than almost any major handgun round and the individual pellet energy (at most about 60 ft-lbs) is barely 20% of almost any major handgun round and only about 1/3 the energy of each shotgun pellet, which he states as the reason the shotgun is so effective. It would seem, if his reckoning were consistent, that a .410 shotshell handgun would only be 1/3rd of 5/9th as effective as a shotgun, or about 18% as effective, which would put it's stop rate at about 15%.
Even worse, when he subtracts the "1/2", he actually only reduces the chart by an unspecified, but much smaller than "1/2" number. There is enough info there to calculate the number though.... he shows it 15% higher than handguns, which are rated at 56%... so that would be 71%.
So, how did reducing the shotgun's 86% by "1/2", result in a 71% rate?
Finally, in what he calls "the most interesting data of all", the "incapacitation success rate" (around 9:50), he points out that there is almost no overall difference in effectiveness between handguns, shotguns and rifles. He then says that "some just do it in 1 or 2 shots instead of 3"... but his chart shows the lowest at 1.22 and the highest at 1.87... in my math, those numbers are all between 1 and 2, which "digitally" means 2... unless we know the Standard Deviation.