I plan to apply a within-subjects design. However, currently, I am investigating my pilot data in which a between-subjects design was applied. The planned model comprises a subject-specific random interaction effect with the factor condition, which is not varied within subjects in the pilot data. I am now asking myself whether this subject-specific interaction is even meaningful if one factor is only varied between subjects or if I should drop this effect for the power analysis to get a more realistic power calculation of another effect of interest? Any help is appreciated.