On Tue, 9 Aug 2005 01:37:54 +0100, Sherif Karama
<[log in to unmask]> wrote:
>Hello,
>
>A few years ago I came across the following design:
>
>ArArArArBrBrBrB
>
>with r being a 'baseline' condition while A and B were 'neutral'
>and 'emotional' conditions, respectively (r=15 sec, A and B = 30 sec
>each). The rationale of the authors was that they wanted to avoid having
>condition A contaminated by condition B (assuming it could happen if a
>block or blocks of B were to be placed before a block of A during a given
>run).
>
>Is there anything wrong with this design (e.g. is it susceptible to low
>frequency noise? Is it sub-optimal for some other reason?).
Find attached some code I wrote showing that A-B is confounded with low-
frequency noise.
>
>Thank you in advance,
>
>Sherif
>=========================================================================
% Original design: ArArArArBrBrBrB
% A, B blocks 30 s each; r blocks 15 s each
fprintf(1,'Original design: ArArArArBrBrBrB\n');
A = [repmat([ones(1,30) zeros(1,15)], 1, 4), zeros(1,165)];
B = [zeros(1,165), repmat([zeros(1,15), ones(1,30)], 1, 4)];
d = B - A;
t = 1:345;
t = spm_detrend(t);
C = corrcoef(d',t');
fprintf(1, 'Correlation coefficient of B - A and a linear trend: ');
fprintf(1, '%.2f\n\n',C(1,2));
% Intermixed design: ArBrArBrArBrArB
fprintf(1,'Intermixed design: ArBrArBrArBrArB\n');
A2 = [repmat([ones(1,30) zeros(1,60)],1,3) ones(1,30) zeros(1,45)];
B2 = [zeros(1,45) ones(1,30) repmat([zeros(1,60) ones(1,30)],1,3)];
d2 = B2 - A2;
C2 = corrcoef(d2',t');
fprintf(1, 'Correlation coefficient of B - A and a linear trend: ');
fprintf(1, '%.2f\n\n',C2(1,2));
|