Dear all, I am implementing a factorial analysis with two within-subjects factors: Factor A with three levels corresponding to different recording sessions of each subject, and factor B with two levels corresponding to measurements within each session. What kind of sphericity assumptions should I make about the data? (Independence and variance settings in full factorial analysis). A further complication is that participants move a lot more in the two A conditions than in the third one, so measurement quality probably differs. For example, should I assume that the levels of both variables are dependent because they are repeated measures from the same subjects? Or should I assume INdependence with A, since the recordings have been made on different sessions? Can I safely assume equal (or similar enough) variance within the levels of factor B? Thank you for any advice. - Jari Kätsyri