What is the outlier criteria in box & whisker plots?

[#1]

Hi everyone,

I´m preparing some box & whisker plots, using the  "outlier" option to exclude outliers from box limits calculation, and showing them as individual points in addition to the boxes. In some cases, some points quite far from the rest are shown as outliers, in other cases are shown as part of the boxes as the "minimum" limit. I wonder which criteria JMP uses to determine wheter a certain point is considered as an outlier or not. A maximum allowed difference or so? If I can get some help on this, It would be great. Thanks in advance!  

[#2 · Accepted solution]

John Tukey developed the box and whiskers plot in the 1970s for a visual data summary based on the five points. The modified box and whiskers plot added visual aids for unusual observations. He deemed any observation more than 1.5 times the inter-quartile distance as "out" and anything more than 4.5 times the IQR as "far out." JMP uses this original criterion in the Outlier Box Plots.

See JMP Help about the Distribution platform reports.

[#3]

Hi Mark!. Thank you for your answer, that´s what I wanted to know. In fact, I have made the verification with some of my JMP boxes I generated with points that seemed to be outliers, and they meet this condition. Thanks a lot!