BookmarkSubscribeSubscribe to RSS Feed

NP chart with variable sample sizes: what is JMP doing?


Community Trekker


Sep 11, 2012

Both the JMP documentation and other references suggest an NP control chart of the number of defective units per sample should have a constant sample size. If I have a variable sample size, I should use a P chart of the proportion defective.

However, if I create an NP chart with a variable sample size, JMP creates a control chart with varying control limits and doesn't chastize me, at least not where I can see.

For example, starting from the Washers sample data set, I add a column titled "Lot Varies" with initial values random, normal distribution, mean = 400, standard deviation = 100, no decimal places. Then I launch Analyze >> Quality & Process >> Control Chart >> NP. In the launch dialog, # Defective is assigned to Process, Lot is assigned to Sample Label, and Lot Varies is assigned to Sample Size. JMP creates a perfectly OK-looking control chart.

Looking at the y-axis, it appears that the units are still in number of defective units, so JMP hasn't decided to develop a P chart instead just because I gave it a variable sample size. But JMP commonly makes reasonable assumptions about what to do based on the data supplied. What is JMP doing here?

Thanks for your assistance.