Please enable JavaScript to view this site.

Prism offers the Newman-Keuls test (sometimes called the Student-Newman-Keuls test) for historical reasons, but we don't recommend ever using it.

This test is a powerful way to compare all pairs of means, reporting statistical significance but not confidence intervals or multiplicity adjusted P values. The problem is it is too powerful, and so it does not maintain the family-wise error rate at the specified level(2). In some cases, the chance of a Type I error can be greater than the alpha level you specified.  

1.MA Seaman, JR Levin and RC Serlin, Psychological Bulletin 110:577-586, 1991.

 

© 1995-2019 GraphPad Software, LLC. All rights reserved.