A Problem With Outlier Tests

When can you really use Dixon’s test?

Outlier tests such as the W-ratio test and Dixon’s outlier test suffer from a problem that can mislead the user. This article will outline the problem and provide guidelines for the appropriate use of these tests.

…

Want to continue?

By logging in you agree to receive communication from Quality Digest. Privacy Policy.

Create a FREE account

Forgot My Password

Comments

Grubbs' test vs. Dixon's test

Dear Sir,

Thank you for sharing such a useful article of outlier test.

Some software use Grubbs's test as one of the options. Which method is more powerful or robust? What is the advantages of using these outlier test since Boxplot could do the job for us?

Regards,

Chee Han LOK

Your Question

Based on my recollection, I would guess that since Grubb's test uses the global standard deviation statistic it has to be less powerful than Dixon's test simply because the outlier will inflate the denominator more with Grubb's test and will deflate the ratio. However, this is only based on my memory, and could be wrong. For more on the usefulness of the Box plot see my Column for April 2014.