 |
 |
Dear Stephen, Thanks for your help on this issue. I think I will indeed opt for taking the average of each twin pair. Dennis -----Oorspronkelijk bericht----- Van: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] Namens Stephen J. Fromm Verzonden: donderdag 4 september 2008 15:17 Aan: [log in to unmask] Onderwerp: Re: [SPM] Estimation of between subjects factor in repeated measures design On Tue, 2 Sep 2008 14:12:16 +0200, Ent D.vant <[log in to unmask]> wrote: >Dear SPMers, > > > >I am studying fMRI in twins that either both have low or high attention >problems. Since measurement of monozygotic twins should be regarded as >correlated I want to enter them as repeated measures and study the >difference in Attention Problems as a between subjects factor. So for >example if I would have 2 twin pairs (pairs 1 and 2) with high attention >problems and 3 twin pairs with low attention problems (pair 3 and 4 and >5), I would for example like to make the following design > > > > > >Pair1: Twin1 vs. Twin2; covariate attention problems =1 > >Pair2: Twin1 vs. Twin2; covariate attention problems =1 > >Pair3: Twin1 vs. Twin2; covariate attention problems =2 > >Pair4: Twin1 vs. Twin2; covariate attention problems =2 > >Pair5: Twin1 vs. Twin2; covariate attention problems =2 > > > >and want to estimate the effect of the covariate which signifies high or >low attention problems. I would be very grateful if anyone could explain >to me how such a 2nd level design could be implemented in SPM5. > > > >Or I would like to be informed on any other alternative approach to >achieve the between twin comparison described above. Be careful with your use of the covariate. The variable you're modeling with the covariate isn't really a "numerical" variable, but is rather a categorical variable. It might be OK in this case, in that you only have two values to the variable. It turns out that in that case, the objection I state above isn't true. Though I think things are clearer if you instead think of the attentional variable as describing groups. So you'd have two groups in this case. As for the repeated measures stuff, you can follow the advice people give on dealing with that (e.g. see questions posted here). Or you could just e.g. take the average of each twin pair, and then enter that into the 2nd level analysis. Best, S > > > >Thank you in advance, Dennis > > > > > >Dr. Dennis van 't Ent > >Dept. Biological Psychology > >Fac. Psychology and Education > >Van der Boechorststraat 1 > >1081 BT, Amsterdam > >The Netherlands > >Tel. +31-20-5982534 > >Fax. +31-20-5988832 > >E-mail: [log in to unmask] > > > >