On Fri, 6 Aug 2004 17:58:00 +0100, Will Penny <[log in to unmask]>
wrote:
>zhang wrote:
>
>> Dear Mr. Penny,
>>
>> I hope I don't bother you too much. But I'm wondering about how to
analyse
>> mixed blocked/event related design.
>>
>> For instance, I will investigate sustained effects in A, B, C blocks,
mean
>> while, these blocks containing x, y events to reveal transient effects.
My
>> purpose is to compare different active patterns of sustained and
transient
>> effects.
>>
>> When I define design matrix in SPM2, may I define A,B,C as trials 1,2,3,
and
>> define x and y as other regressors, or vice versa? Or analyse these two
>> effects separately?
>>
>>
>
>
>You can define A, B and C as blocks (ie. duration > 0) and x and y as
>events (duration=0).
>
>The only problem I foresee is that your block effects may be very
>collinear with your event effects eg. if block A comprises lots of x events
>these two regressors would look very similar - effectively you'd be
modelling the
>same thing twice. Other than that your model looks OK.
>
>Alternatively, you could forget about using events and model the
>transient nature of responses with a parametric modulation
>by time ie. have your block regressors and also other block regressors
>but with an exponential decay during each block (or something of that
>sort).
>
>Also, remember that it will not meaningful to compare events and
>blocks because they are scaled differently when the design matrices are
>created. So, only compare events to events and blocks to blocks.
>
>Best wishes,
>
>Will.
>
>
>>
>> Thank you very much!
>>
>>
>>
>> Best regards,
>>
>>
>>
>>
>>
>> Zhang De Xuan
>>
>> ---------------------
>>
>> Center for Brain and Cognitive Sciences
>>
>> Department of Psychology
>>
>> Peking University
>>
>> Beijing 100871
>>
>> China
>>
>> Fax: +86 10 6276 1081
>>
>> Tel: +86 10 6276 8875
>>
>>
>>
>>
>
>
>--
>William D. Penny
>Wellcome Department of Imaging Neuroscience
>University College London
>12 Queen Square
>London WC1N 3BG
>
>Tel: 020 7833 7478
>FAX: 020 7813 1420
>Email: [log in to unmask]
>URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|