 |
 |
I don't know of the solution and I did a quick search of the archives, but didn't find any solutions. However, I would recommend that you do the analysis in a statistically valid way. If you are set on using the eye(N) to define your contrast, I would suggest moving to GLM_flex where you can properly specify the error term as the between-subject effect. An equally valid approach would be test whether any of the FIR bins are different from the others with the contrast: 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 0 0 0 0 0 0 0 0 1 -1 Best Regards, Donald McLaren ================= D.G. McLaren, Ph.D. Postdoctoral Research Fellow, GRECC, Bedford VA Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School 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 Mon, Mar 26, 2012 at 6:00 AM, Bianca de Haan < [log in to unmask]> wrote: > Hi Donald, > > Thank you very much for your rapid reply. > > I do take your point, that as I'm not comparing the within-subject levels > to one another, one might consider it akin to performing multiple > one-sample t-tests, so it's strictly speaking not really a within-subject > effect that I'm testing... GLM_flex sounds like an interesting toolbox and > I will give it a try. The idea of the eye contrast was derived from the SPM > manual (chapter 30: face group fmri data). As in the example in the manual, > my factor is time and the 9 levels are the FIR time bins. I basically would > like to assess those voxels that show any form of event-related response, > analogous to the example in the manual. > > The problem remains though that the same-vein contrast (looking where any > of the levels of a within-subject repeated measures factor are different > from 0) works with other datasets, so I'm still not sure why it would fail > to run on this particular dataset and would appreciate any hints or > suggestions. > > I apologize in advance if I missed something obvious, > > Bianca > > > On Fri, Mar 23, 2012 at 6:00 PM, MCLAREN, Donald <[log in to unmask] > > wrote: > >> Bianca, >> >> The contrast you plan use (eye(n)) to test whether any of the values are >> different than 0 is not a valid contrast as this would be a between-subject >> effect and the error term is for within-subject effects. If you use >> GLM_flex, then you can evaluate the contrast with the appropriate error >> term. >> >> See below for answers to your other issues. >> >> On Fri, Mar 23, 2012 at 12:40 PM, Bianca de Haan < >> [log in to unmask]> wrote: >> >>> Dear all, >>> >>> When estimating a repeated measures 2nd level model in the latest >>> version of spm8 (r4667, however, the problem remains when switching back to >>> r4290) running on Matlab R2010b, I keep getting the following error >>> message: >>> >>> Failed 'Model estimation' >>> Error using ==> eig >>> Input to EIG must not contain NaN or Inf. >>> In file "C:\Programme\spm8_R4667\spm_reml.m" (v3791), function >>> "spm_reml" at line 202. >>> In file "C:\Programme\spm8_R4667\spm_spm.m" (v4515), function "spm_spm" >>> at line 875. >>> In file "C:\Programme\spm8_R4667\config\spm_run_fmri_est.m" (v4403), >>> function "spm_run_fmri_est" at line 33. >>> >>> The following modules did not run: >>> Failed: Model estimation >>> >>> >>> This error message is preceded by multiple warnings of "Matrix is >>> singular, close to singular or badly scaled. Results may be inaccurate. >>> RCOND = NaN." >>> >>> Similar problems have been reported on the list ( >>> http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind1202&L=SPM&D=0&I=-3&d=No+Match%3BMatch%3BMatches&P=705417and >>> http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0807&L=SPM&P=R7435&I=-3&d=No+Match%3BMatch%3BMatches), >>> however, my batch does not contain duplicate filenames and changing the >>> spm_reml.m file around line 78 did not solve the problem. >>> >>> My design is a flexible factorial with factor 'subject' (independence >>> 'yes', variance 'unequal') and my repeated measures factor with 9 levels >>> (independence 'no', variance 'equal'). Under 'main effects and >>> interactions' I specified only the main effect of my repeated measures >>> factor (factor number '2'). >>> >>> Ultimately, I would simply like to test where any of my levels on the >>> repeated measures factor is significantly different from 0 (i.e. eye(9) >>> contrast) >>> >> >> See above comment about this being invalid. >> >> >>> >>> The error remains when I set the variance of the subject factor to >>> 'equal'. However, when additionally specifying the main effect of the >>> subject factor (factor number '1') under 'main effects and interactions' >>> the estimation runs without a problem. >>> >> >> You need to include the main effect of subject in the model to properly >> do a repeated-measures analysis. >> >> >>> >>> I encounter no problems when performing similar analyses on other >>> datasets, which leads me to suspect that it must be something specific to >>> this particular dataset... The results of preprocessing and the 1st level >>> analyses, however, look fine. >>> >>> I am at a loss how to solve this problem and would be grateful for any >>> hints or advice that might help me. >>> >>> Many thanks in advance, >>> >>> Bianca >>> >> >> > > > -- > Dr. Bianca de Haan > Center of Neurology, Division of Neuropsychology > University of Tübingen > Hoppe-Seyler-Str. 3, D-72076 Tübingen > +49 (0)7071 29 85661 > http://homepages.uni-tuebingen.de/klinikum/bianca.de-haan/ >