Jws, obviously i agree that vivs method was flawed and required far to many assumptions but in the end the umber was near identical.
as for taking the repeat rate of sexual predators into acount, the 19 per 1000 includes that by default. Its a flat rate of total reports no matter who the offender was.
I agree its plausible to extrapolate that world wide but im nowhere near as confident. There may well be different rates of offense based on social norms, reporting law may skew data sets etc...