s-news
[Top] [All Lists]

Dunnett's pvalue

To: <s-news@lists.biostat.wustl.edu>
Subject: Dunnett's pvalue
From: "Zhan, Ping N." <Ping.Zhan@celera.com>
Date: Thu, 24 Feb 2005 11:34:44 -0500
Thread-index: AcUahJKLsVqC7MlDSeu6SARk4h9JmwACEFSA
Thread-topic: Dunnett's pvalue
Hi group,

Please allow me to follow up on this question, whose original posting
was attached. For not being a mathematician, could someone show me the
steps of implementing the inverse of qdunnett in Splus. It does not look
trivia to me. Thanks in advance.

Ping Zhan

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~
On Mon, 2 Dec 2002, Peter Christen wrote:

> Does anyone have an idea or a solution for calculating p-values for
the
> Dunnett test?
>
> For the t-distribution, S-PLUS provides the functions qt and pt, where
> qt(p,df) calculates quantiles for a given probability and pt(q,df) the
> cumulative probability for a given quantile.
>
> Similarly, for Dunnett's test, the function qdunnett calculates
> quantiles for a given significance level. Unfortunately, the inverse
> function for calculating p-values is not available.

You can write a function to invert qdunnett via uniroot.  You want to
solve  t = qdunnett(p, ...)  for p, where t is the observed value of the
statistic (if I understand you correctly).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~


<Prev in Thread] Current Thread [Next in Thread>
  • Dunnett's pvalue, Zhan, Ping N. <=