I know this is an "armchair academic" comment, but I wonder if `places that are likely to have active surveillance` strongly intersects with `places that the bystander effect is less likely`. In other words, I wonder if there's a crippling selection bias.

The reason I think this is possible is because I'm way more likely to help someone if I don't think I'm endangering myself by doing so. Areas that are actively surveilled are probably more likely to be safe areas for me to do so.