STATS MAKE ME CRY

I'm examining a repeated measures MANCOVA with two outcomes (T1 and T2 for each as repeated measures) with 5 covariates. I request the estimated marginal means output and plot to see what the means are for T1 and T2 adjusting for the covariates. My primary multivariate and between subjects output shows a significant effect for change in Variable 1 and for change in Variable 2. However, the Estimated Marginal Means output shows an increase from T1 to T2 for Variable 2 (which is NOT a good thing and NOT what we expect) - there is an additional "multivariate" output table accompanying each EMM table for each outcome. THAT table shows that differences between T1 and T2 for Variable 2 are NOT significant - and clearly based on the very small difference, this appears to be an accurate statistic.

I am confused about why the primary multivariate output and the EMM multivariate output do not match. Aren't the statistics in the primary (and by primary, I mean the overall multivariate output with all covariates and such at the beginning of the output) supposed to be accounting for the effects of the covariates on the outcome (change/time for Variables 1 and 2)?

I cannot locate any annotated description of the SPSS output that includes these EMM multivariate tables online or in any of my SPSS books.

Thank you for your time!

Best,
Chelsea