Date:         Wed, 14 Mar 2007 17:30:47 -0800
Reply-To:     David L Cassell <davidlcassell@MSN.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         David L Cassell <davidlcassell@MSN.COM>
Subject:      Re: bimodal
In-Reply-To:  <45F7A789020000C900011C21@MAIL.NDRI.ORG>
Content-Type: text/plain; format=flowed

flom@NDRI.ORG replied:
>
>johbee@GMAIL.COM wrote:
> >
> >Does anyone know if there is a way to detect bi or multi modal
> >distributions using surveyselect, or univariate, or mean?
>
>
>
>There is a dip test invented by Hartigan.  (google those three words, get
>lots of refs), and it seems to be available in SAS (add SAS to the google
>search above.....but I am away from the office today, and don't have time
>to check which of what Google supplies is good.....
>
>
>
>HTH
>
>Peter

Hartigan's dip test really tests unimodal against 'other', although it is
traditionally
used against a bimodal alternative.  It won't test bimodal against all
alternatives
(unimodal, trimodal, ...).

There's an R package that will do Hartigan's dip test for you.  The R
package
has a much more robust set of p-values, based on a much larger simulation
set than Hartigan & Hartigan used back in 1985.

And there's also a test by Cheng and Hall (1999, Annals of Statistics).

But the bottom line is: if you can see it, you can find it with the test; if
you
can't be sure with your eyes, you can't be sure with the test either.
That's
a rule that holds for an awful lot of stat tests, not just this one.

HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330

_________________________________________________________________
Play Flexicon: the crossword game that feeds your brain. PLAY now for FREE.
  http://zone.msn.com/en/flexicon/default.htm?icid=flexicon_hmtagline