Dear David,
The second model you suggest here would be the one that I would use as it puts the information in the model which to me makes more sense, although I believe that they would both give equivalent stats when taken to the third level.
All the best,
Mark
On 2 Feb 2013, at 22:33, David Soto <[log in to unmask]> wrote:
> hi - think this is a pretty basic question
> so apologies for that
>
> We are interested in modelling a learning effect
> in a simple desigm
>
> Each participant has 2 runs, 5 blocks each
>
> For each block we modelled EVs for predictive and unpredictive cues, leading to 10 EVs
>
> For the first-level analyses, we computed 5 copes for the difference
> between predictive > unpredictive cues
>
> We are interested in the learning effect, namely how the
> difference between predictive and unpredictive cues changes across
> the 10 blocks
>
>
> We then submit, for each subject, the 10 copes from the 2 first level-analyses
> into a Fixed effects analyses
>
> specifying the design as
>
> 1 0 0 0 0 0 0 0 0 0 0
> 0 1 0 0 0 0 0 0 0 0 0
> 0 0 1 0 0 0 0 0 0 0 0
> 0 0 0 1 0 0 0 0 0 0 0
> 0 0 0 0 1 0 0 0 0 0 0
> 0 0 0 0 0 1 0 0 0 0 0
> 0 0 0 0 0 0 1 0 0 0 0
> 0 0 0 0 0 0 0 1 0 0 0
> 0 0 0 0 0 0 0 0 1 0 0
> 0 0 0 0 0 0 0 0 0 1 0
> 0 0 0 0 0 0 0 0 0 0 1
>
> and (for example) a linear contrast
> -4.5 -3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5 4.5
>
>
> I guess this model look fine?
>
> or would it be more appropriate to do a Fixed effects with one EV for the relevant cope
> and another covariate to model the learning effect, such as
>
> 1 -4.5
> 1 -3.5
> 1 -2.5
> 1 -1.5
> 1 -0.5
> 1 0.5
> 1 1.5
> 1 2.5
> 1 3.5
> 1 4.5
>
> and then assess the covariate effect for each subject and bring that to the 3rd level
>
> which model would be more appropiate in this case? guess the 2 models are pretty much
> the same..what is the difference if any
>
> or is there a better way to model for this experiment?
>
> suggestions welcome and appreciated!
>
> thanks so much
>
> David
|