Normal and nonparametric methods for tolerance intervals
main topics
When you calculate tolerance intervals
using Stat > Quality Tools
> Tolerance Intervals, Minitab displays results for both the normal
and nonparametric methods. Minitab also displays results for both methods
when you plan for tolerance intervals using Stat
> Power and Sample Size > Sample Size for Tolerance Intervals.
To determine which method to use, consider the following:
· If
your data follow a normal distribution, then the normal method is more
precise and economical than the nonparametric method. The normal method
allows you to achieve smaller margins of error [link] with fewer observations.
Use the normal method if you know from prior experience or analysis that
your population is normally distributed. A normality
test,
such as the one that is included with Stat > Quality Tools > Tolerance
Intervals, can also help you decide.
· The
normal method is not robust to severe departures from normality. If you
are unsure of the parent distribution, or you know that the parent distribution
is not normal, then use the nonparametric method. The nonparametric method
requires only that the data are continuous.
· The
nonparametric method requires large sample sizes to cover a large percentage
of the population. If your sample size is not large enough, the achieved
confidence level for your tolerance interval can be much lower than the
desired level.
Minitab uses an exact method to calculate normal tolerance intervals
(see [6, 8, 9]). Minitab's
sample size planning for normal tolerance intervals also uses an exact
method (see [3, 4, 6, 8,
9, 10, 16]).