Hello,

i'm currently running an analysis of longitudinal data. i'm using HLM analysis using the SAS software.

the design is as follow:

IV1 - time (3 time points: before\during\after stress) - coded as a categorical variable.

IV2 - brain activity (3 time points: before\during after stress)

DV - questionaire score (3 time points: before]during\after stress).

covariance structure: VC

no random effect of time (based on model fit indices comparing random vs. non-random effects)

in the time as the only predictor model there I've found a significant change in the DV between the first and second time-point but to change between the second and third time point.

I want to explore how brain activity reflects this this pattern of change. (i.e - how brain activity change between time 1 to time 2 and whether it changes between time 2 to 3).

I've tried to center my IV2 according to baseline (given it represent the period before the stressful event).

doing so resulted in an IV2 that represent deviation from baseline - so at basline it equals to "0" (zero) and in the other time point is equal to deviation from baseline.

coding the variable in this way makes it imposibble to search for the different between time point 1 to the other time points (beacuse the simple effect of this time point will not exist).

however, using mean-centering doesn't make any since given my design. is it more reasonable to use deviation from time 2 as the refrence point? i want to see the change in slopes between the time-points.

i will be happy for any kind of insight regarding how to approach this analysis.

thank you from advance,

Noa