Doug-

Thanks for the info- it was very helpful. One last question. Since I am planning on concatenating the three runs, would it be better for me to create a psuedo-randomization process where half of the trials for each condition (A, B,C, and D) begin with the TR and the other half begin at the half TR? Given your previous response, could the unequal N due to randomization explain the higher/lower standard deviations between the even and the odd coefficients for each condition (note that here I am not talking about the correlations within even and odd but rather the descrepencies between the two)? If this is the case wouldn't it be better to put a constraint on the randomization rather than have such differing estimates for the same trial type?

PS- there was some suspicious activity during the meteor showers recently. However, some guy named Fox and another two guys in suits who were wearing sunglasses even though it was night assured me that it was ok. the incident becomes hazier and hazier as time passes ........