Repeated measures ANOVA (EEMs vs descriptive means)

Hi all, I hope someone can help me to clarify one issue in my results. I´m running mixed ANOVAs in SPSS 22 with LL(slef-control) as measure, one fixed within subject factor (3 different probabilities) and one between subject factor (Question) that changes in each ANOVA runned (but it is always 2 categories: Ideal and Not ideal).

In the post-hoc results, the means of the probabilities (within subject factor) are different compared to the means showed in the Estimates Table. I can´t understand why the mean for the within subject factor changes from one table to another because the between subject factor has not been introduced in the analysis yet.


Cause the between subject factor changes in each ANOVA runned, the n for each group mean are not equaly (i don´t now if this has an influence in this issue).

Why is it happening? Am I doing something wrong?