A mixture of normals can have just about any shape (see the book Finite
Mixture Models by McLachlan and Peel, which has, near the beginning, a
small bestiary of distributions generated with mixtures of normals.); it
need not be normal, or even remotely close to normal, if you allow the
means and variances to vary.

It could be unimodal, even if the means differ.  One way this could
happen is if the means are relatively close, compared to the variances. 


HTH

Peter

Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
www.peterflom.com
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)


>>> [EMAIL PROTECTED] 05/03/04 1:29 PM >>>
Although apparently "a mixture of two normals, differing in means, can
still
be unimodal" (orig. Everitt, 1981 "Bimodality and the nature of
depression") -- which I don't quite 'see' yet.

----- Original Message ----- 
From: "rd3d" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Monday, May 03, 2004 10:41 AM
Subject: [edstat] Mixture of normal distributions


> I would like to know if the mixture of normal distributions is also
> considered a normal distribution (with mean and variance of the
mixture as
> parameters).
> If one looks at the graph of the mixture, it is not really
bell-shaped, so
> I am wondering if it can still be normally distributed.
>
> .
> .
> =================================================================
> Instructions for joining and leaving this list, remarks about the
> problem of INAPPROPRIATE MESSAGES, and archives are available at:
> .                  http://jse.stat.ncsu.edu/                    .
> =================================================================
>

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to