Dear All,
Following on from the recent thread, I tried using the
efficiency calculation
trace(c'*inv(xX.X'*xX.X)*c)^-1 (Friston et al, 99)
to compare the efficiency of different design matrices
using the same contrast [1;0] and 100 scans (TR=3). I'm
slightly confused by these results:
---------------------- --------------------------------
Event-Related design Blocked design
Single regressor Single regressor
(HRF only - fixed SOA) (Unconvolved boxcar - fixed SOA)
---------------------- --------------------------------
SOA Efficiency SOA Length Efficiency
--- ---------- --- ------ ----------
0.5 136.04 14 7 24.99
1 29.9365 28 14 24.64
2 6.4547 42 21 24.36
3 44.8916 56 28 24.64
4 77.2743 100 50 25
8 98.3636
16 61.9545
32 37.8401
64 19.9191
Are they in keeping with conventional wisdom?
Any ideas as to what I might have done wrong?
Yours,
Afraim Salek-Haddadi
Clinical Research Fellow
Institute of Neurology
|