A relevant, classic bystander-effect experiment (Darley and Latane 1968) put several participants, all subjects of the experiment, in a waiting room. Smoke began to come into the room through a vent. In most groups, no one acted for a very long time, even after the smoke led to breathing problems for participants. By contrast, when only one person was in the room, not a group, that person would almost always leave the room or call for assistance.

Differences from the OPs (very interesting) study: 1. In the smoke room, participants know they are in an experimental setting, though they do not know that the waiting room is part of the experiment. 2. The entire group is visible in the waiting room. (In a crowd outside, you don’t know who else might be acting or watching. In the waiting room you can see that no one else is doing anything.) 3. The threat is to you personally as well as to the others in the group - an injury or a fight involving others is less personally affecting than smoke that is making you cough in an enclosed room. 4. D & L experimented with small groups including a single person and saw dramatically different results in the latter case. I’d be interested in seeing what happens late at night when there are just one or two people around - what do the camera show then?

Of these I suspect 2) has the greatest effect. If I see that literally no one in the room is doing anything, I may figure that the situation is probably under control or isn’t a big deal. In a big crowd outdoors I can’t draw that conclusion as clearly.