To be fair, ambushing anyone on the street, out of context, is extremely likely to get a dumber answer than normal.

This would make a good study, actually. Ambush people on the street and quiz them. For the control group, quiz them immediately, on the street. For the experimental group, take them to a more relaxed location, allow them to sit, give them a minute or two to get collected, quiz them and measure the difference from the control group. You can do it with easy, medium, hard questions.