Overview for Tolerance Intervals (Nonnormal Distribution)

Tolerance intervals are a range of values for a specific quality characteristic of a product that likely covers a specified minimum percentage of current or future product output. Use Tolerance Intervals (Nonnormal Distribution) to determine the range within which a specified minimum proportion of the process measurements is expected to fall when your data come from a specified nonnormal distribution.

You can create two-sided or one-sided intervals depending on what you want to determine, as follows:

Use a two-sided interval to determine the interval that contains a certain minimum percentage of the process measurements.
Use a one-sided interval to determine the interval such that a certain minimum percentage of the process measurements will not be less than a lower limit, or will not be greater than an upper limit.

For example, a manufacturer wants to evaluate the variability of their paper bleaching process. The brightness of the pulp varies from one batch to the next. The manufacturer uses a two-sided 95% tolerance interval to determine the range within which the brightness of at least 99% of the batches are expected to fall.

Where to find this analysis

To perform a tolerance interval analysis, choose Stat > Quality Tools > Tolerance Intervals (Nonnormal Distribution).

When to use an alternate analysis

If you want to calculate tolerance intervals when your data come from a normal distribution, use normal tolerance intervals.