Hello,
I have some questions regarding event related design for a task examining
effects of stimulus repetition,
Our parameters are as follows: TR = 2, stimulus durations = 2, ISI = 2. We
use 10 stimuli per run, repeated 5 times each (4 runs total, each with
discrete stimuli sets).
I used Optseq2 to calculate optimal randomization of stimuli and for
generating variable length null events.
I have 3 questions:
1. Regarding the amount of null spacing: I have read that the more null
events one includes in a randomized design, the more sensitive one will be
to detecting main effects, however more null events also reduces sensitivity
to differential effects. As I see it, we would like to examine repetition
effects using a parametric regressor in the design matrices to examine
linear increases and decreases of activity across each run, however we would
also like to construct design matrices that could examine differential
effects between, e.g. the 1st and 5th repetitions. So I'm fairly certain I
want a number of null events that equates approxiamtely equal efficiency for
main effects and differential effects contrasts. Hmmm..
2a. What approximately would this number be given that per run I have
essentialy 1 condition of 10 (similar) stimuli, each of which is repeated 5
times? (In Optseq2, I treated the stimuli as 10 distinct conditions,
repeated 5 times apiece, and allotted an amount of null time equivalent to
the each of the 10 conditions)
2b. Because my null events vary in length from 2-10 s, should I simply be
thinking about equating total amount of null time with my condition (as
oppose to number of null events)?
Thanks in advance for any help!
Taylor
|