So where do you draw the line? Please tell me the precise location where a woman can be assured of zero risk when encountering a stranger.
Is this a realistic question? Zero risk is your standard?
How do you leave the house each day?
Now, had you asked about reasonable or relative risk when encountering a stranger, I suggest you base your decisions/actions on real risk, and try to avoid perceived risk when the perception is inconsistent with real risk.
My philosophy is based on the fact that there is no such thing as zero risk, and avoiding risk comes at a cost. The easy things are a no brainer. Make sure you have a working smoke alarm in your house, and that you check the batteries a couple times a year. Wear a seatbelt in any vehicle. Look both ways when you cross the street.
Some things are more costly. A Volvo has excellent crash safety, but it is very expensive. There is evidence some diets are healthier than others but it takes a lot of effort for some people to stick to a healthy diet. Beyond the easy no brainer risk avoidance measures we need to make decisions that we find acceptable. If that Volvo is important to you, you spend the money. If you can't, then you buy the safest vehicle you can afford. If an unsafe sports car is something you really want, then base the decision on real data about the safety, not on impressions of safety or lack of it. If you love bacon and are willing to trade your indulgence for the cardiac risk, same thing, base the decision on real data, not on perceived risk.
The bottom line is we make risk avoidance decisions all the time. But most people make those decisions based on perceived risk rather than real risk. And studies of perceived risk vs real risk reveal there are a number of things which distort perceived risk for most humans. False beliefs about the risk to women at 4am in an elevator in a 5 star hotel in a 'foreign country' for example, is not surprising given what we know about how people perceive risk. Rational thinkers would try to match those perceptions with the actual data (real risk) as much as possible.