 |
 |
If you want to do an analysis in which you look for a main effect of IQ across all subjects, then you should also demean across all subjects -- not separately demean the controls and patients. See Jeanette Mumford's web page. cheers, -MH On Wed, 2011-11-23 at 09:40 -0500, Roger Jou wrote: > Hi Dave, > > Thanks so much for your reply. I just wanted double check that I have this > right. Below I have IQ scores in the left column and demeaned IQ > scores in the > right column. The first 11 rows are controls and the remaining 19 rows are > patients. For the demeaned values for controls and patients, I > calculated this > by [individual control IQ – control mean IQ] and [individual patient IQ – > patient mean IQ], respectively. > > 106 -8.4 > 127 12.6 > 131 16.6 > 114 0.1 > 117 2.6 > 92 -22.4 > 113 -1.4 > 123 8.6 > 111 -3.4 > 110 -4.4 > 114 -.4 > 127.0 35.7 > 107 15.7 > 102 10.7 > 100 8.7 > 86 -5.3 > 51 -40.3 > 114 22.7 > 108 16.7 > 80 -11.3 > 76 -15.3 > 78 -13.3 > 90 -1.3 > 85 -6.3 > 100 8.7 > 93 1.7 > 121 29.7 > 84 -7.3 > 41 -50.3 > 91 0.1 > > If this is correct, then are the followng design.con and design.mat files > correct? > > design.con file: > > /NumWaves 3 > /NumContrasts 2 > /PPheights 1 1 > /Matrix > 1 -1 0 > -1 1 0 > > design.mat file: > > /NumWaves 3 > /NumPoints 30 > /PPheights 86 86 > /Matrix > 1 0 -8.4 > 1 0 12.6 > 1 0 16.6 > 1 0 0.1 > 1 0 2.6 > 1 0 -22.4 > 1 0 -1.4 > 1 0 8.6 > 1 0 -3.4 > 1 0 -4.4 > 1 0 -0.4 > 0 1 35.7 > 0 1 15.7 > 0 1 10.7 > 0 1 8.7 > 0 1 -5.3 > 0 1 -40.3 > 0 1 22.7 > 0 1 16.7 > 0 1 -11.3 > 0 1 -15.3 > 0 1 -13.3 > 0 1 -1.3 > 0 1 -6.3 > 0 1 8.7 > 0 1 1.7 > 0 1 29.7 > 0 1 -7.3 > 0 1 -50.3 > 0 1 0.1 > > Also, I wanted to double check that the PPheights are the difference > between the > most negative and most positive demeaned IQ for the entire sample. > > Finally, in order to run this would the procedure be the same as that > indicated > on the main TBSS documentation page, except using the above for the for the > design.mat and design.con files below: > > randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d > design.mat > –t design.con -n 1000 --T2 –V > > Thanks so much your kind assistance! > > Roger > > > Quoting David Flitney <[log in to unmask]>: > > > Simply put your raw IQ scores into one column, demean as one, and > > drop the last column in your contrasts list. This will allow you to > > model the linear effect of IQ(?) on your expected signal... but > > spread the misfit across your signal in a unpredictable way if this > > is not the case. > > > > -- > > Dave Flitney > > IT Manager, FMRIB Centre > > > > On 22 Nov 2011, at 17:37, Roger Jou <[log in to unmask]> wrote: > > > >> Dear Prof Smith, > >> > >> Thank you for clarifying my error. I understand what this means > >> conceptually, > >> but could please show me what the design.con and design.mat looks > >> like? I want > >> to be certain I am on the right track. Thanks again. > >> > >> Regards, > >> > >> Roger > >> > >> Quoting Stephen Smith <[log in to unmask]>: > >> > >>> Hi - no this is incorrect - if you want to "control for" IQ this > >>> needs to be a single EV (across both groups) that will attempt to > >>> model out the effect of IQ. > >>> However that will only succeed (ie remove this confound in the > >>> group difference) to the extent that the effect of IQ on the data > >>> is linear and additive - which cannot necessarily be assumed. > >>> Cheers. > >>> > >>> > >>> On 19 Sep 2011, at 19:42, Roger Jou wrote: > >>> > >>>> Dear FSL Experts, > >>>> > >>>> I am conducting a TBSS analysis on FA, MD, AD and RD and plan to > >>>> use TFCE. The study consists of two groups (control = 11 and > >>>> patient = 19) who differ significantly in IQ. Therefore, I wanted > >>>> conduct the analysis controlling for IQ and would be extremely > >>>> grateful for help in generating the appropriate design.con and > >>>> design.mat files. > >>>> > >>>> I have looked through the forum for answers and have drafted the > >>>> following: > >>>> > >>>> design.con > >>>> > >>>> /NumWaves 4 > >>>> /NumContrasts 2 > >>>> /PPheights 1 1 > >>>> /Matrix > >>>> 1 -1 0 0 > >>>> -1 1 0 0 > >>>> > >>>> Can you please tell me whether this design.con file is correct? > >>>> > >>>> design.mat > >>>> > >>>> /NumWaves 4 > >>>> /NumPoints 30 > >>>> /PPheights 86 86 > >>>> /Matrix > >>>> 1 0 -8.4 0 > >>>> 1 0 12.6 0 > >>>> 1 0 16.6 0 > >>>> 1 0 0 0 > >>>> 1 0 2.6 0 > >>>> 1 0 -22.4 0 > >>>> 1 0 -1.4 0 > >>>> 1 0 8.6 0 > >>>> 1 0 -3.4 0 > >>>> 1 0 -4.4 0 > >>>> 1 0 -0.4 0 > >>>> 0 1 0 35.7 > >>>> 0 1 0 15.7 > >>>> 0 1 0 10.7 > >>>> 0 1 0 8.7 > >>>> 0 1 0 -5.3 > >>>> 0 1 0 -40.3 > >>>> 0 1 0 22.7 > >>>> 0 1 0 16.7 > >>>> 0 1 0 -11.3 > >>>> 0 1 0 -15.3 > >>>> 0 1 0 -13.3 > >>>> 0 1 0 -1.3 > >>>> 0 1 0 -6.3 > >>>> 0 1 0 8.7 > >>>> 0 1 0 1.7 > >>>> 0 1 0 29.7 > >>>> 0 1 0 -7.3 > >>>> 0 1 0 -50.3 > >>>> 0 1 0 0 > >>>> > >>>> Can you please tell me whether this design.mat file is correct? > >>>> > >>>> Also, I wanted to double check that the PPheights are the > >>>> difference between the most negative and most positive demeaned IQ > >>>> for the entire sample, and the IQ needs to be demeaned separately > >>>> for control and patient groups. > >>>> > >>>> Finally, in order to run this would the procedure the same as that > >>>> indicated on the main TBSS documentation page, just using the > >>>> above for the for the design.mat and design.con files below: > >>>> > >>>> randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask > >>>> -d design.mat -t design.con -n 500 --T2 -V > >>>> > >>>> Thanks in advance for your help! > >>>> > >>>> Roger > >>>> > >>> > >>> > >>> --------------------------------------------------------------------------- > >>> Stephen M. Smith, Professor of Biomedical Engineering > >>> Associate Director, Oxford University FMRIB Centre > >>> > >>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK > >>> +44 (0) 1865 222726 (fax 222717) > >>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve > >>> --------------------------------------------------------------------------- > >>> > >>> > >>> > >>> > >> > >