a minor point: i think that, by most standards, between 1% and 5% is far from being astonishingly precise, because these numbers define a range. the precision of a range is measured in its (relative) length, not by the precision of its endpoints.
put it this way, if i said an event has a...