If you established a set of criteria which gave you a 'serious risk of collision' and then ran the experiment 67 times, noting that said collision never happened...would you change the criteria? To something slightly less sensational such as 'nearer miss'...
I have noted a risk of collision on countless occasions, I've taken them all relatively seriously.