The interaction of MC and switching was clearly significant, F (1, 39) = 25.018, p < .001, η 2 p = .391, confirming larger error switch costs in the modality-incompatible than in the modality-compatible condition (6.5% vs. 3.5%).
The two-way interaction of MC and S-R compatibility was non-significant too, F (1, 39) = 3.128, p = .085, η 2 p = .074, with only a numerical trend for a larger effect of S-R compatibility in the modality-incompatible condition (94 ms) than in the modality-compatible condition (73 ms). 2 Bayesian analysis The follow-up Bayesian repeated-measures ANOVA yielded, for RT, moderate evidence for H 0 concerning both the interaction of MC and S-R compatibility (BF 10 = 0.23) as well as concerning the three-way interaction of MC, S-R compatibility, and switching (BF 10 = 0.25).