I have a 2 (temporal orientation: future versus present, but treated
as a continuous variable in the analysis) x 3 (treatment condition: no
intervention, treatment 1, treatment 2) design and a continuous DV
(measured at 2 time points: pre-intervention and post-intervention). I
am not quite sure which of the following two methods is more
appropriate: (1) using difference scores, posttest - pretest, as DV or
(2) using the post-intervention data as DV and controlling for pretest
data (including it as a covariate). When regressing CFC and condition
on the pre-intervention data, I only get a main effect for CFC but not
condition. Also the pre-intervention and post-intervention data itself
are highly correlated.