Okay, issues with the clip. Unlike Ian, I'm not phased by the sole usage of wire-frame body language as an indicator, if that is the aim of the experiment (although I do, of course, agree that there are a large number of other factors involved). The issues instead lie in the set up of the entire thing, from the selection of the viewers to the "solution" provided.
The first guys are bouncers, so they're used to not looking for victims, but for those that will start trouble. Regardless, they didn't do too bad (the gender bias I found interesting... when I watched the second wire-frame, my first thought was that it was a girl. These guys noticed the hip movement, and said "this guys got a swagger, almost feminine hip movement"... come on, guys, there was no indication it was a guy!), and were easily accurate in their selection of the assault victim. Good, fair enough, nice and easy. So far so good, we might think... but we'll come back to that.
The next scene is the "training", where the victim was put through an almost RBSD seminar, including getting him to strike things hard, and correcting what they could about his awareness and body language, over 4 hours. They pushed him past a few social boundaries, and had what they would probably consider a fair amount of success. That was followed by having David (the victim) again filmed with motion capture reflectors, and to then have that clip shown to a new group of viewers, this time a group of boxers, with 7 out of the 10 picking the 4th person, and the other three picking David.
Which is where it all falls apart. Completely.
The boxers, like the bouncers, are not street fighters, predators, or criminals. They are, instead, boxers. I know, shocking, huh? What does that mean? Well, it means that what they look for in terms of someone to fight isn't going to be the same thing that a criminal/street predator would. And that showed completely in their choices.
They chose what were actually the two strongest body language persons as their only choices, the two that showed the most confidence in their stride. Whether that was because they didn't want to be seen as picking on weaker persons (which doesn't lead to any respect in boxing, as you are matched up before a fight, and there is a lot of pride in a fight being a good one, rather than a one-sided affair), or just due to an ingrained preference that they have, they, to a man, shied away from picking number one, who was clearly the weakest of the lot, or number two, clearly a female, and typically therefore seen as an easier target. So their testimony can be thrown out straight away. Additionally, how is the vast difference in the number of viewers accounted for? Two out of two picked David the first time, and three out of ten the second time... what if we had 10 bouncers, would six of them pick the girl? If we only chose two of the boxers, would they be the ones who picked David? Was there any real thought given to such a fundamental flaw in the experiment?
But what about the change we saw in young David? Surely that was a good indication of the effect that body language can have on people watching you. Yeah, okay. Except that there was no guarantee that the change was in any way permanent, or even long-lived. Nor is there any indication of what his body language was before the assault. It was most likely markedly different, as assault can have quite severe psychological effects, including, but not limited to, causing a person to be more fearful in their daily life, which has a direct effect on their body language. In other words, the body language that the bouncers saw, causing them to pick David as the "weakest" might have been the result of the assault, not the cause of it.
Then let's look at the timetable of the end of the experiment. David is taken off to his four hour training session, and then filmed again straight afterwards (Lee says "David, an unusual afternoon for you.... before we finish up, I want to do one more thing, and that is to have your movement motion captured again"). What that means is that all the training, and it's effects, are uppermost in David's mind, including the endorphin rush he would now be feeling after the adrenaline of the training session itself, which would act as a "pick me up", and increase his body language by itself. My honest bet is that, if we were to film David again in another six months, he'd be back where he started the clip. His psychology just wasn't that changed in that little time.
So, to sum up...
No establishment of the actual cause of the assault making the premise flawed.
Poor selection of panels to view the wire-frame making the experiment flawed.
Poor execution of a follow-up, making the experiment flawed.
Lack of acknowledgement of other influence, making the experiment flawed by having no true control aspect.
Lack of continuity in the screening and panel, making the experiment flawed.
All in all, a rather flawed, although interesting, clip.