RE: Sample size

[morphmet]

Thought I posted this earlier. Apparently not. Sorry. -mod

-------- Original Message --------
Subject: RE: Sample size
Date: Fri, 13 Nov 2009 08:32:10 -0800 (PST)
From: F. James Rohlf <ro...@life.bio.sunysb.edu>
Organization: Stony Brook University
To: morphmet@morphometrics.org
References: <4afd72f9.1040...@morphometrics.org>

There is no rule of thumb. Unequal sample sizes are just not an 
efficient use of one's efforts in making measurements.

For example, with two samples of sizes n1 and n2 and n1+n2 = 100 a split
like 10 and 90 has an effective average sample size of just 18 
(equivalent to a total of just 36 samples rather than the actual 100). A 
split of 2 and 98 has an effective average sample size of just 3.92 
(equivalent to a total sample size of just 5.92).

The same considerations apply to multivariate data except that one must 
also be concerned with having sample sizes large enough so that the 
covariance matrix is not singular (see earlier discussion on morphmet).

----------------------
F. James Rohlf, Distinguished Professor
Dept. Ecology and Evolution, Stony Brook University, NY 11794-5245

-----Original Message-----
From: morphmet [mailto:morphmet_modera...@morphometrics.org]
Sent: Friday, November 13, 2009 9:54 AM
To: morphmet
Subject: RE: Sample size



-------- Original Message --------
Subject:        RE: Sample size
Date:   Fri, 13 Nov 2009 04:34:31 -0800 (PST)
From:   Samuel Okoye <samu...@yahoo.com>
To:     morphmet@morphometrics.org



Thank you very much for your reply.

If it is not required to have equal sample sizes (n1=n2), is there any
rule of thump how big the difference between the two group (e.g. n1=5
n2=15)? Does this also apply for multivariate data?

Many thaks in advance,
Samuel

--- On *Thu, 11/12/09, morphmet
/<morphmet_modera...@morphometrics.org>/* wrote:


     From: morphmet <morphmet_modera...@morphometrics.org>
     Subject: RE: Sample size
     To: "morphmet" <morphmet@morphometrics.org>
     Date: Thursday, November 12, 2009, 10:58 AM



     -------- Original Message --------
     Subject: RE: Sample size
     Date: Thu, 12 Nov 2009 10:39:35 -0800 (PST)
     From: F. James Rohlf <ro...@life.bio.sunysb.edu
     </mc/compose?to=ro...@life.bio.sunysb.edu>>
     Organization: Ecology & Evolution
     To: <morphmet@morphometrics.org
     </mc/compose?to=morph...@morphometrics.org>>
     References: <4af5219d.1070...@morphometrics.org
     </mc/compose?to=4af5219d.1070...@morphometrics.org>>

     Do you mean for morphometric data (multivariate) or just univariate
     data? For the univariate case you could check Section 9.8 in Biometry
     for a method to estimate sample size. It does not, however, consider
     the unequal sample size case.

     In general, equal sample sizes are not required. The main advantage of
     equal sample sizes is that they are more efficient. For a given total
     sample size, n1+n2, you will have greater statistical power if n1=n2.
     Having equal sample sizes also make the computations slightly simpler
     but that is no longer much of an issue when computers are used to
     perform all of the calculations.

     =========================
     F. James Rohlf
     Distinguished Professor, Stony Brook University
     http://life.bio.sunysb.edu/ee/rohlf


      > -----Original Message-----
      > From: morphmet [mailto:morphmet_modera...@morphometrics.org
     </mc/compose?to=morphmet_modera...@morphometrics.org>]
      > Sent: Saturday, November 07, 2009 2:28 AM
      > To: morphmet
      > Subject: Sample size
      >
      >
      >
      > -------- Original Message --------
      > Subject:     Sample size
      > Date:     Wed, 4 Nov 2009 06:40:15 -0800 (PST)
      > From:     Samuel Okoye <samu...@yahoo.com
     </mc/compose?to=samu...@yahoo.com>>
      > To:     morphmet_modera...@morphometrics.org
     </mc/compose?to=morphmet_modera...@morphometrics.org>
      >
      >
      >
      > Dear all,
      >
      > I would be most grateful if you tell me how important to have
      > balanced
      > sample sizes for the t-test and ANOVA? Should the sample sizes
      > always be
      > equal? If not how big should the difference between them be? How
      > can I
      > do sample size calculaction for unbalanced t-test or ANOVA?
      >
      > Many thanks in advance,
      > Samuel
      >
      >
      >
      > --
      > Replies will be sent to the list.
      > For more information visit http://www.morphometrics.org



     -- Replies will be sent to the list.
     For more information visit http://www.morphometrics.org



--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org



--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org