Hi All,
What are people's views on using covariates based on differences in higher-level repeated measures analyses?
Specifically, given N higher-level pair-wise contrasts, one could include covariates based on differences in, for example, some behavioural measure (demeaned or normalized). E.g., for the contrast Cond1 - Cond2, one could include Behav1 - Behav2, resulting in maps that are "sensitive" to how differences in "Behave" relate to differences in "Cond".
Thanks!,
Andy
|