Note also, that these p-values are for 1-sided outlier tests, whereas outlier
testing is inherently 2-sided, unless there is some objective basis for
believing that outliers can only occur in one direction. Thus for most
cases, these p-values should be doubled, as in Rorabacher's tables
(Analytical Chemistry 63[2]:139-146, 1991) and USP <111.

Jerry

"Mike Middleton" wrote:

Kevin -

Referring to

http://www.statistics4u.info/fundsta...ts_dixon.html#

it's a simple matter to compute the test statistic Q using Excel's ABS, MIN,
MAX, and LARGE or SMALL worksheet functions.

But I don't have access to a source that describes how to compute the
sampling distribution of Q for determining a p-value or conducting a
hypothesis test.

- Mike
http://www.mikemiddleton.com

"Kevin Clark" <Kevin wrote in message
...
I am looking for a way to automate the Dixon Test for determining outliers
in
a data set in an Excel spreadsheet.