The Robustness of Two-Sided Tolerance Limits for Normal Distributions

The robustness of two-sided tolerance limits for normal distributions is examined based on a computer simulation in which the Student's t and gamma distributions are used as generating models. Results indicate that the normal tolerance limits are sensitive to departures from normality when k-factors are selected for coverages in excess of 0.9. Reasonable robustness is achieved for coverages of 0.9, as long as the underlying distribution is not extremely heavy-tailed nor highly skewed.