Thanks Donald for your help. One final question - putting the "does the amount of variance explained vary by group?" question aside, if I wanted to run a multiple regression style model within one group looking at the contribution of the two behavioral variables together, does it make sense to weight the two variables as part of the same chunk of variance e.g. 0 0 0 0 .125 .125 .125 .125 .125 .125 .125 .125 (which I think will run a model looking at the contribution of variables 1 & 2 together regardless of group or gender). If so, I'm assuming I could adjust the model and run separate GLMs within each group, as you suggested - I just want to make sure I am understanding the weights correctly first...
LT
On Fri, Mar 22, 2013 at 8:23 AM, MCLAREN, Donald mclaren.donald@gmail.comwrote:
On Thu, Mar 21, 2013 at 8:59 PM, Laura M. Tully < tully.laura@googlemail.com> wrote:
oops sorry! (I also miscalculated the # of regressors - there's actually 12 (not 10 as previously noted). Here is the list of column labels: Grp1male Grp1female Grp2male Grp2female Grp1maleVar1 Grp1femaleVar1 Grp1maleVar2 Grp1femaleVar2 Grp2maleVar1 Grp2femaleVar1 Grp2maleVar2 Grp2femaleVar2
And what I think is actually an F test looking for group x var 1 interaction OR group x variable 2 interaction whilst accounting for gender. .5 .5 -.5 -.5 0 0 0 0 0 0 0 0 0 0 0 0 .5 .5 -.5 -.5
The Contrast for group*var1 would be: 0 0 0 0 .5 .5 0 0 -.5 -.5 0 0 The Contrast for group*var2 would be: 0 0 0 0 0 0 .5 .5 0 0 -.5 -.5
But what I actually WANT to test is a multiple regression style model - i.e. if I put var 1 AND 2 into the model together do they explain more variance than either variable alone, AND does this vary by group (is this even a sensible contrast to make?). Which I *think* would look something like this...
0 0 0 0 .125 .125 .125 .125 .125 .125 .125 .125
People generally don't ask that question. The answer is when you add more variables, you will explain more variance. Tests about overall model fits are generally assessed with the AIC, BIC, etc. metrics. I'm not sure if there is anyway in regression to say that the amount of variance explained is different by group unless you run 2 separate models. If you think this might be a valid question, I'd consult a statistician - which I am not.
Laura.
On Thu, Mar 21, 2013 at 5:51 PM, MCLAREN, Donald < mclaren.donald@gmail.com> wrote:
Please include the list of the column labels.
Best Regards, Donald McLaren
D.G. McLaren, Ph.D. Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School Postdoctoral Research Fellow, GRECC, Bedford VA Website: http://www.martinos.org/~mclaren Office: (773) 406-2464 ===================== This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is intended only for the use of the individual or entity named above. If the reader of the e-mail is not the intended recipient or the employee or agent responsible for delivering it to the intended recipient, you are hereby notified that you are in possession of confidential and privileged information. Any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this e-mail unintentionally, please immediately notify the sender via telephone at (773) 406-2464 or email.
On Thu, Mar 21, 2013 at 6:43 PM, Laura M. Tully < tully.laura@googlemail.com> wrote:
hi Experts,
I'm struggling to conceptualize the appropriate contrasts for my cortical thickness analysis. I have four classes [two groups; two levels (patients,controls; male,female) and two behavioral variables. I want to see if together the two variables account significant proportion of the variance in y (thickness) and if this differs by group whilst regressing out gender. - i.e. if I enter both behavioral variables into the model does it account for more variance than either variable on their own (after controlling for gender)? What I have is this:
.5 .5 -.5 -.5 0 0 0 0 0 0 0 0 .5 .5 -.5 -.5
Does this look right?
Thanks!
Laura.
--
Laura M. Tully, MA Social Neuroscience & Psychopathology, Harvard University Center for the Assessment and Prevention of Prodromal States, UCLA Semel Institute of Neuroscience ltully@mednet.ucla.edu ltully@fas.harvard.edu 310-267-0170 -- My musings as a young clinical scientist: http://theclinicalbrain.blogspot.com/ Follow me on Twitter: @tully_laura
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
--
Laura M. Tully, MA Social Neuroscience & Psychopathology, Harvard University Center for the Assessment and Prevention of Prodromal States, UCLA Semel Institute of Neuroscience ltully@mednet.ucla.edu ltully@fas.harvard.edu 310-267-0170 -- My musings as a young clinical scientist: http://theclinicalbrain.blogspot.com/ Follow me on Twitter: @tully_laura