I feel that NBC's study was definitely flawed. The child in the video was not a good actress and did not act legitimately terrified. She didn't start crying, kicking, hitting, or even screaming at the top of her lungs. I really did not get a sense of fear from her which I'm sure has to be hard to create when you're not really in danger. She didn't act any different from a disobedient child throwing a tantrum. Since everyone has seen hundreds of tantrums but most people have never witnessed an actual child abduction, it's understandable that people would assume that it was a child just being a brat to her stepdad.
An abductor who decides to abduct a random child is most likely going to do it in front of as few witnesses as possible. If they do decide to abduct a child from a downtown area, they would most likely use lures to get the child to come into their car. Grabbing a screaming child and throwing them into your car with dozens of witnesses around is extremely risky. The only case I can think of where that occurred was Samantha Runnion and there was only one witness and thanks to her description, he was caught within a couple of days of Samantha's body being found. Etan Patz was kidnapped from a busy NYC street yet no one saw a thing. Why? Because the abductor most likely got Etan to go into his car willingly by using questions/lures.
The best thing for anyone in a scenario like the one shown in the video to do would be to single out one person for help. Look them straight in the eye and tell them to call 911 and get you help. That puts all the responsibility on that one person and it will make it more likely that they wouldn't depend on someone else to do something.
There are 72 million children under the age of 18 in the United States. 115 children a year are kidnapped every year by strangers. There is so much paranoia over something that has a 0.000159722222% chance of happening to the average child. I think there is a big correlation between stranger danger and child obesity.