This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: inverse values for the F distribution
- From: Rajarshi Guha <rajarshi at presidency dot com>
- To: Martin Jansche <jansche at ling dot ohio-state dot edu>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: 02 Sep 2003 17:08:23 -0400
- Subject: Re: inverse values for the F distribution
- Organization:
- References: <Pine.SOL.4.33.0309021531400.27048-100000@julius.ling.ohio-state.edu>
- Reply-to: rajarshi at presidency dot com
On Tue, 2003-09-02 at 15:40, Martin Jansche wrote:
> On 2 Sep 2003, Rajarshi Guha wrote:
>
> > In the abscence of this function is there any way to get a
> > critical value for an F test statistic?
>
> If you absolutely need a critical value, you could use a root finding
> algorithm. But are you sure you need one? If you're doing a simple F
> test, it might be enough to compute the upper tail probability using
> gsl_cdf_fdist_Q() and then compare that probability with your desired
> significance level.
Thanks. Actually I was doing the latter but the values I was getting
seemed surprising (or maybe I just dont know what types of values to
expect!). For example gsl_cdf_fdist_Q(3.4935,3,28) gives me 4223.0 which
is clearly greater than 0.05 (my desired level of significance). I had
thought of checking by calculating a critical value as well.
Just out of curioisity, why is there no inverse function for the F
distribution?
-------------------------------------------------------------------
Rajarshi Guha <rajarshi@presidency.com> <http://jijo.cjb.net>
GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE
-------------------------------------------------------------------
Q: What do you get when you put a spinning flywheel in a casket and
turn a corner?
A: A funeral precession.