Thanks Steve! What do you exactly mean by very few repeats? I have 4
trials per EV if I split them. Does it seem to you very few?
On 3/5/09, Steve Smith <[log in to unmask]> wrote:
> Hi,
>
> Yes unless you have very few repeats of each condition I would
> recommend one EV per conditions - i.e. 4 EVs. This will give you the
> greatest flexibility in modelling and possible contrasts, and (in
> general) will lead to higher zstats (even including for contrasts not
> involving these conditions) as it will reduce the residuals in the
> model fitting, when compared with using just 2 EVs.
>
> Cheers.
>
>
> On 5 Mar 2009, at 21:20, Shan Luo wrote:
>
>> Hi FSLers, I have a 2 ( win/loss) X 2 ( high/low) design. Now I have
>> two ways to code my
>> events.
>>
>> One way is to code the 2 high ( high win and high loss) and 2 low
>> (low win and low loss)
>> each as one event and then just do a simple contrast between the
>> two. The other is to code
>> the 2 high as 2 separate events and the 2 low as separate events and
>> to then do the
>> contrast as the 2 high events - the 2 low events. These appear to
>> yield different results --
>> is one preferable / more powerful than the other?
>>
>> The other question is: will the choice between the 2 alternatives
>> above affect statistics for
>> contrasts that don't involve those events?
>>
>> Thanks very much.
>>
>> Shan
>>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
>
--
Shan (Hannah) Luo
PhD Student
Department of Psychology
University of Southern California
Work phone: 213-740-1357
|