Also not knowing the context, I answered it as referring to what standard jurors are instructed to apply. Then I would have selected 1.0-5.0 % , but had to settle for 0.1%. Of course, in science, I'd set a tighter standard, still not 0, more like what judges say is "beyond any conceivable doubt"
It depends. 0.001 probability in ~10^6 events, gives You 10^3 events if population is large enough...
It depends on the stakes. I would require a significantly lower chance of being wrong to convict someone of murder than to determine that someone was cheating at poker, for example.
Not knowing the context, I picked the smallest number available. But I'd actually take the term to mean that an average person would reasonably doubt something, which in terms of numbers would mean that at least 50% - 60% of the population was 100% certain. Maybe...