Hello everybody,

I have a question about the selection of specific comparisons on SPSS while running a GLM or GLMM. The background of the database is the following: 

• the dependent variable is the sum of occurrences of a behavior in birds
• 2 treatments called ENC and CON, with 4 units/each (1,2,3,4)
• 5 moments of the day (A, B, C, D, E)
• 3 weeks of age (17, 32, 44)

I considered the treatment, moment of day, and weeks as fixed factors, and included the unit as repeated measurement. 

I noticed that if I run the statistics considering this factors and I just ask for the output of triple interaction week*treatment*moment I run out od degrees of freedom and I do not see significant differences e.g. between C CON-A CON even when numbers are really different. I know this because the standard deviations on the graph are really far from each other and it is impossible that no significant differences are detected. In confirmation of this, if I use a database with just e.g. C CON-A CON at the 3 weeks of age I do see the differences instead.

This problem could be solved if I could select just the comparisons that I need within the full database. For example, I do not care to compare A-CON at 17 weeks to B-CON at 32 weeks, as it would be like comparing apples with bananas. At the same time, if I compare everything with everything this banana-apple comparison happens in default. My question is, how can I select this with SPSS? I guess it is something related to syntax and probably the program screen I shared below, but I really need help to solve this, especially beacuse I need the triple interaction but in the window I do not see any option other than "pairwise comparisons". I apologize for the picture, my SPSS is italian!

Thank you in advance for your help!!