3 ms·

That's interesting, because it is quite improbable that someone would "know personally" several lottery winners at ~590000 to 1 odds, even if his acquaintances

That's interesting, because it is quite improbable that someone would "know personally" several lottery winners at ~590000 to 1 odds, even if his acquaintances routinely bought two tickets a day. GP's post is just a single datum, of course, but it's enough in itself to raise doubt about the accuracy of that 1.7/megatrip statistic. Numerically, suppose GP knows 30 people who make 600 trips on BART each year. In one year, the chance of at least one robbery in that group, given the quoted odds, is three percent. The chance of multiple robberies, i.e. at least two, is under 0.05 percent – over 2000 to 1 odds against. Even over a ten year period, the odds are 25 to 1 against. If GP's personal experience is not unusual, then something is screwy with the stat.

thaumasiotes 9y ago
> GP's post is just a single datum, of course, but it's enough in itself to raise doubt about the accuracy of that 1.7/megatrip statistic. In itself it's not that great. You have two obvious options: 1. Robberies on BART are generally not reported. You'd need a very high rate of nonreporting to mess with the conclusion that overall BART trips are quite safe. 2. Certain BART trips, by location, time of day, or some other metric, are much more dangerous than the average, and the BART-using public preferentially avoids taking those trips. If you believe in case 2, it's not at all surprising that someone might know several people who have been robbed on BART. You'd conclude that their social circle has atypical BART habits. How many robberies-at-gunpoint do you think occur in the packed rush hour trains?