The homogeneity of slope test is really just a test of the covariate by group interaction. So in a 2x2x2 ANCOVA you could test up to three covariate by group interactions and you might also want to test the covariate by interaction interactions also. Generally one wants to keep all individual difference factor by covariate interactions even if not significant (though you need a reasonable size of data set to make this practical).

Thank you for your answer Thom S Baguley.

I just want to specify my factorial plan, my covariate is Vocabulary scores because it is different between older and younger adults (my between factor) and because exprimental design is with words, i believe that this factor, a posteriori, could have an influence on my dependant variable (response time). To test homogeneity of slope, I have to create a personalized (customized ?) plan (e.g., tutorial on ANCOVA with SPSS), and i would like to know if i have to enter all effects (age, vocabulary, age x vocabulary) or just interaction between my between factor and covariate ?

Thank you,

Marcellin

Personally I hate the SPSS interface and when I do these analyses in SPSS I tend to manually create the interaction terms I want to add (after centering the covariate etc.).

You need to include all the terms below the interaction but my recollection is that SPSS lets you do that with A * B shorthand or by A + B + A x B (but I could be wrong).