>>39
I stand corrected, I did mean translation-invariant rather than just non-compactly supported.
Anyway, to clarify my point, let me suggest alternate formulations of the OP question:
1/ Consider a group of 5000 people, one in 100 of whom wears glasses. Pick one at random (uniformly): what's the probability of her wearing glasses?
2/ Consider a group of N>5000 people, one in 100 of whom wears glasses. Pick 5000 of them at random (uniformly): what's the expected probability of one those wearing glasses?
3/ Consider a group of N>5000 people, one in 100 of whom wears glasses. Pick 5000 of them: what's the probability of one those wearing glasses?
4/ I belong to a group of N>5000 people, one in 100 of whom wears glasses. Pick 5000 of them at random, myself included: what's the expected probability of one those wearing glasses?
1 and 2 have 1/100 as an obvious and definite answer. You can't say anything about 3. And the answer to 4 depends on whether "I" wear glasses or not. That's only a few possible variations.