RANOVA using unbalanced design

Can anyone please advise if there is a way to use ranova with an unbalanced within-subject design? I've tried using NaNs for instances where I don't have a measurement, but that returns an error. Thanks.

[Johannes Rebling]

Same question here. Anyone?

[Scott MacKenzie]

I'd like to clarify what you mean by "unbalanced within-subjects design"? Let's say your experiment engages 10 participants were are tested on one factor with three levels. By "unbalanced" are you saying that some of the participants were not tested on all three levels? That breaks a fundamental tenet of within-subjects -- that each participant is tested on all levels of a factor. If this is the situation you are in, perhaps you should recruit new participants to replace the errant ones.