JiscMail Logo
Email discussion lists for the UK Education and Research communities

Help for FSL Archives


FSL Archives

FSL Archives


FSL@JISCMAIL.AC.UK


View:

Message:

[

First

|

Previous

|

Next

|

Last

]

By Topic:

[

First

|

Previous

|

Next

|

Last

]

By Author:

[

First

|

Previous

|

Next

|

Last

]

Font:

Proportional Font

LISTSERV Archives

LISTSERV Archives

FSL Home

FSL Home

FSL  June 2007

FSL June 2007

Options

Subscribe or Unsubscribe

Subscribe or Unsubscribe

Log In

Log In

Get Password

Get Password

Subject:

Re: Randomise multiple regression NON orthogonal EVs and confound

From:

Jaroslav Hlinka <[log in to unmask]>

Reply-To:

FSL - FMRIB's Software Library <[log in to unmask]>

Date:

Thu, 28 Jun 2007 20:19:38 +0100

Content-Type:

text/plain

Parts/Attachments:

Parts/Attachments

text/plain (36 lines)

Dear Steve,

Thanks again for the advice given so far.
I have another question about Randomise analysis of TBSS data
I have two confounds, non-orthogonal to the EV of interest. I understand 
that I should orthogonalise the EV wrt BOTH confounds, but am not sure, 
how exactly to do it.


For beginning, I tried using the orthogonalise button in Glm Gui for one 
EV and on confound, in the .mat file I get variables which are orthogonal 
(dot product equals 0), but NOT AT ALL uncorrelated (I hypothesise their 
won't be unless at least one of the EVs entered had mean zero!). I guess 
that is not what I want, is it? Therefore in my data situation (1EV, 2 
confounds) I HAVE TO pre-demean EV of interest (and the confounds as 
well?), use Glm gui, copy the orthogonalised EV of interest from .mat to 
new Glm matrix and the confounds to separate Glm matrix (!cannot use 
the .mat as it is, since confounds have to go separately for –x option!). 
Is that right?

Is it ORTHOGONALISATION which is necessary, or is “DECORRELATION” enough? 

By decorrelation I mean computing the (partial in case of multiple 
confounds) regression coefficients (RC) for EV dependent of confound, and 
substracting RC*confound from the EV. 

Or, since computing partial regression coefficients is not 
straightforward, I can first orthogonalise (decorrelate) confound 1to wrt 
confound 2 and then orthogonalise EV wrt confound 1 and the orthogonalised 
EV orthogonalise again wrt confound 2. Is that right?

I hope I expressed clearly what is my problem and what are my hypothesis 
for solution…

Thanks for any help,
Jaroslav 

Top of Message | Previous Page | Permalink

JISCMail Tools


RSS Feeds and Sharing


Advanced Options


Archives

April 2014
March 2014
February 2014
January 2014
December 2013
November 2013
October 2013
September 2013
August 2013
July 2013
June 2013
May 2013
April 2013
March 2013
February 2013
January 2013
December 2012
November 2012
October 2012
September 2012
August 2012
July 2012
June 2012
May 2012
April 2012
March 2012
February 2012
January 2012
December 2011
November 2011
October 2011
September 2011
August 2011
July 2011
June 2011
May 2011
April 2011
March 2011
February 2011
January 2011
December 2010
November 2010
October 2010
September 2010
August 2010
July 2010
June 2010
May 2010
April 2010
March 2010
February 2010
January 2010
December 2009
November 2009
October 2009
September 2009
August 2009
July 2009
June 2009
May 2009
April 2009
March 2009
February 2009
January 2009
December 2008
November 2008
October 2008
September 2008
August 2008
July 2008
June 2008
May 2008
April 2008
March 2008
February 2008
January 2008
December 2007
November 2007
October 2007
September 2007
August 2007
July 2007
June 2007
May 2007
April 2007
March 2007
February 2007
January 2007
2006
2005
2004
2003
2002
2001


WWW.JISCMAIL.AC.UK

Secured by F-Secure Anti-Virus CataList Email List Search Powered by the LISTSERV Email List Manager