 |
 |
Hi Juan, Go with the 1st one -- i.e., no repeated data in two models, but rather a single model for all data used only once. All the best, Anderson On 27 June 2016 at 12:10, Juan Pablo Martin Trias <[log in to unmask]> wrote: > Dear all! > > > I'm modeling an event-related memory recognition task. > > In this fMRI model we have 3 kind of trials: > > -NEW PICTURES-->50% of all inputs. > > -PICTURES corresponding to the Vertex Condition --> 25% (pictures that > during the encoding were presented with Vertex repetitive Transcranial > Magnetic Stimulation). > > -PICTURES corresponding to the DLPFC Condition --> 25% (pictures that > during the encoding were presented with Dorsolateral Prefrontal Cortex > repetitive Transcranial Magnetic Stimulation). > > > For all of these events we have correct answers (hits and correct > rejections) and wrong answers (misses and false alarms). And, of course the > onset of all the events. > > > We have then 12 subjects with fMRI onsets for Hits in two categories > (Vertex and DLPFC). > > My question is: Which one of the next models is better fitted to the data? > > 1) 12 inputs and 4 explanatory variables : Hits Vertex, Hits DLPFC, Misses > Vertex, Misses DLPFC. This model let me do contrasts with hits vs misses > activations or interacions. > > 2) 2 separate models of 24 inputs with Hits or Misses. The first 12 inputs > are related with Vertex condition and the next 12 for the same 12 subjects > in the DLPFC condition, like in a repeated measures model. In this model i > can add an EV for each 12 subjects to control for each subject activation. > I can't contrast in the same model Hits and Misses. > > I'd really appreciate your help!!! > > > Many thanks > > > > >