Help needed with interpretation of a (clear) 2x3-ANOVA interaction

#1
Dear Stats community

I hope someone can help me interpreting my results.

I ran two tests, each with three participant groups. So I have a within-subject factor with 2 levels and a between-subject factor with 3 levels. Interaction is significant.

I attached a picture to show you the data.

How do I proceed from here? Which group caused the interaction?
From the picture I infer, that the red group caused the interaction. I ran three additional repeated-measures-anova for red-green, red-blue, and green-blue comparisons. These tests verified my guess. But I think this is not a legal statistical procedure. How can I statistically find out which of the three groups caused the interaction?
Is it even possible?


I am thankful for any idea or help.
Have a great day

minimax




**************************************************************
additional information

The x-axis of the picture just shows test 1 and 2. The y-axis shows scores. The lines are the connected performances of the three different groups for test 1 and 2.
In case you can't see the picture: I have three groups: red, green and blue.
red and blue line cross each other, so that red scores higher in test 1 and blue scores higher in test 2. the green line is parallel to the blue one, but on overall lower scores and no crossing with the red line.


First I ran a 2x3 repeated measures anova. Both main effects (groups and tests) were significant. Interaction was significant, too.

Afterwards I ran simple main effects analysis:
For all groups the scores in test 1 and test 2 are significantly different.
For test 2 the scores of the different groups are not significantly different.
For test 1 the scores of the different groups are significantly different.

multiple comparison (bonferroni corrected+) for test 1 showed that the green group scores significantly different from the other two.
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Seems like you made the right approaches. As can be seen the slopes are not congruent. Did you have an interaction term in your model? What program are you using?
 
#3
Hey hlsmith

thanks for the reply. Sorry, but I am not quite sure what you mean by interaction term. I only know it from regression models.

I used SPSS to run a two-way-repeated-measure-ANOVA with a 3 level between-subjects factor and a 2 level within subjects factor. And interaction was significant.
I know I am repeating myself, but maybe you can explain what you mean by interaction term :)

Thank you very much,
minimax
 
#4
I just want to refresh the treat and still hope for some help.
I still don't know exactly what was meant with the interaction term by hlsmith.

Thanks for any hints.

Best
minimax