Skip to content

[Fwd: effective sample size]

4 messages · David Atkins, John Maindonald, Rafael Maia +1 more

#
This was just posted to the multilevel listserv, and I have taken a very 
quick glance.  It appears to offer an alternative form for estimating 
degrees of freedom.  It's focused on longitudinal data, and I have not 
looked closely enough to see whether it would be more generally 
applicable to mixed-models (multilevel, nested, crossed, etc.) or 
whether it would "scale up" for larger problems.

Given all the hoo-hah around dfs, thought I would at least kick it out 
to the group.

cheers, Dave

-------- Original Message --------
Subject: effective sample size
Date: Tue, 17 Nov 2009 09:12:47 -0600
From: Stas Kolenikov <skolenik at GMAIL.COM>
Reply-To: Multilevel modelling discussion list <MULTILEVEL at JISCMAIL.AC.UK>
To: MULTILEVEL at JISCMAIL.AC.UK

The recent issue of The American Statistician contains a pretty neat
paper on the effective sample size and degrees of freedom in
longitudinal studies: see
http://www.citeulike.org/user/ctacmo/article/6129798. Highly
recommended; there aren't that many people who understand longitudinal
data as well as the Geert duo.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

-------------------------- Multilevel list --------------------------
To leave, send    leave multilevel    to jiscmail at jiscmail.ac.uk
For further info about the Multilevel list, please see
http://www.jiscmail.ac.uk/lists/multilevel.html     and
http://www.nursing.teaching.man.ac.uk/staff/mcampbell/multilevel.html
2 days later
#
That is a neat paper.  The formula that is used for calculating
effective sample size is remarkably simple, an obvious thing
to try once one has seen it.  It seems to me likely that, by taking
better account of off-diagonal elements in the variance-covariance
matrix, it should be possible to improve on their formula.

John Maindonald             email: john.maindonald at anu.edu.au
phone : +61 2 (6125)3473    fax  : +61 2(6125)5549
Centre for Mathematics & Its Applications, Room 1194,
John Dedman Mathematical Sciences Building (Building 27)
Australian National University, Canberra ACT 0200.
http://www.maths.anu.edu.au/~johnm
On 18/11/2009, at 3:10 AM, David Atkins wrote:

            
#
Could somebody possibly send me a copy of the pdf? I don't have access to that journal at this time...

thanks!


Abra?os,
Rafael Maia
---
webpage: http://gozips.uakron.edu/~rm72
"A little learning is a dangerous thing; drink deep, or taste not the Pierian spring." (A. Pope)
Graduate Student - Integrated Bioscience
University of Akron
http://gozips.uakron.edu/~shawkey/
On Nov 17, 2009, at 11:10 AM, David Atkins wrote:

            
#
David Atkins wrote:

            
...
(That's G.Moolenberghs and G.Verbeke, for the lazy and uninitiated.)

The notion that the information content can be bounded as the number of 
data points go to infinity should come as no shock to anyone who has 
looked at experimental data with subsecond resolution, BTW. At the end 
of the day, you still have only five rats!