In every single news story that I can remember about someone getting stranded with a broken-down car, the person leaves their car looking for help and they end up dead. Or, if there are multiple people in the car, the one person who goes off looking for help ends up dead and the other people who stayed with the car survive.
My question: If I get stranded in my car somewhere, should I go looking for help? Or should I follow my availability heuristic and stay put? Since I'm still in the process of debiasing myself, should I do the opposite of what this particular mental shortcut suggests, or would that be a sort of bias bias (analogous to the fallacy fallacy)?
Don't decide to randomly explore until you find help. If you know where you're going for that help, that's an entirely different thing.
Desert is brutal.
And extreme cold is worse.
I think the selection bias here is massive. Every story where the person dies will get reported, and every story where the person is fine will not. I would look for help.
This is true whether the person stayed with their car or looked for help. There is a bias here, but it should make both strategies seem less effective. Since the problem is a lack of normalization, the bias will be stronger for whichever strategy is used more often.
I haven't done the estimate, but if there are several people stranded, it seems to make sense to split up to improve the overall chances of survival if the odds of being found are unreliable (all eggs in one basket and such).
Please also note the selection bias, where the happy ending stories (the person who left the car quickly found and brought back help) do not get nearly as much publicity. Presumably there are public stats available, if anyone here can find them.
Well, I'd say a good place to start thinking about this might be to recall the circumstances that led to those people dying, and use that to inform your estimate of how dangerous a particular situation might be. For example, if you're mired in a snowdrift in mountain country during the winter, with a blizzard on the way, staying put is probably the better bet. If you're stuck in an irrigation ditch in farmland during the late spring, on a sunny day, looking for help is likely safe and may -- depending on where you are -- be faster. Being slightly paranoid about unfamiliar territory (like desert) is probably a good idea.
As a general policy, I think I'd be likely to go look for help if and only if I knew I had a good chance of getting somewhere I was significantly more likely to find that help, taking local conditions into account. Keep in mind that you're probably a lot more likely to successfully flag down a passing car if you're standing next to a disabled vehicle at the time -- and especially so if the vehicle's got some obvious damage.
Additional tips for flagging down vehicles
Be a well-endowed female
Have a visible injury
Look them in the eyes
I just thought of a number of female anime characters who sport an eyepatch.
There are some situations in which you should probably leave your car:
1) If staying in your car leaves you in immediate danger (you don't want to be stuck in a flash flood)
2) You have a specific destination in mind that you can walk to and probably get help.
Otherwise, being in the car is probably better than not.
Two highly relevant pieces of information that would effect my decision:
How high is the crime rate in the area?
Are you armed (and if so, to what extent)?
EDIT: Downvoters, would the fact that you are in Mission Viejo, California or Baltimore, Maryland not effect your preferred strategy or is there something else about this comment that makes it of low value?
Didn't downvote, but "stranded" makes me think 'middle of the desert' and doesn't at all make me think 'bad neighborhood', so maybe they thought your comment was non-topical.
I'm also curious.
Perhaps there should be a web API equivalent to 911? Maps apps seem to be a good place to put communications services for rescue. This would have to be supported using per-app asymmetric keys or white-box encryption. (To foil vandals.)
I'd also be able to view a history of travel-related fatalities in a Maps app.