Models for Computing Upper Error Limits in Dollar-Unit Sampling