JMP Wish List
We want to hear your ideas for improving JMP. Share them here.What inspired this wish list request? In JMP 20 EA, there is an option for saving tolerance intervals from Fit Mixed. Like other intervals in the general Fit Model platform, all intervals are two-sided. When a one-sided interval is desired for confidence and prediction intervals, you can simply adjust the alpha value to get the inference (e.g., a lower one-sided 95% prediction interval can use alpha=0.10 and only report the lower PI value). However, tolerance intervals use different formulas based on one and two-sided cases. Hence, a specific option when saving tolerance intervals in the pop up to choose one-sided lower or one-sided upper is desired.
What is the improvement you would like to see? When choosing the tolerance interval option, you could have a pop up similar to what is shown below for the Distribution platform (first figure). The only thing you would delete is the Distribution choices. Currently we get a familiar pop up (second figure), but just need the radio buttons for two-sided (default), and the one-sided options.
Why is this idea important? Tolerance intervals have similar value to confidence and prediction intervals. In many cases, critical attributes may only have one side that is critical, such as one with a single specification limit. Hence, only the one-sided interval is critical for the inference. The stability analysis option in Fit Mixed allows for tolerance intervals in a special case model, so there is a precedent for this functionality in those cases. Other software, such as Stat-Ease offer one-sided equations for tolerance intervals as well (Equation 5 - Using DOE with Tolerance Intervals to Verify Specifications).