Post hoc test for Trend analysis and other

#1
Hi all,

I'm new to trend analysis and would like to get some advice.

My study is about treatment intervention on learning.
Between-subject variable is treatment group (treatment1, treatment2, no treatment) and within-subject variable is time (time1, time2, time3). My dependent variable is subject's task score on a behavioural test. I want to compare treatment effects on learning of the task.

First, I did two-way mixed ANOVA (aka split plot ANOVA). The main effect of time was significant but the interaction of treatment and time was not significant.

Next, trend analysis (or Within-subject contrast on SPSS) gave me good results;
Linear trend was significant and the interaction of linear trend was also significant.
If my understanding is correct, the significant interaction of the linear trend means that a linear trend of each group differs from at least one of the other groups..., right?

I would like to dig in this trend analysis because I'm interested in how learning curves/trends differ across the treatment groups.

Here are my questions;
1) Is there any post hoc test for trend analysis to see which treatment group differs from which treatment group?

2) How do you calculate the df of F value for the interaction of the linear trend? (I know this is a stupid question but I cannot find any good instruction.)

3) I'm not completely convinced by the fact that mixed ANOVA didn't show a significant interaction while linear trend interaction was significant. Can anyone provide me with any possible reason for this?

I use mainly R and SPSS and sometimes SAS, but I'm open to other softwares and manual analysis.

Any advice would be greatly appreciated.
Thank you in advance!