My recollection is that this number is based on a percentage of individuals seeking paternity tests-- where obviously we would expect to see a much larger number of false paternities showing up than in the general population.