A multi-sample Shivplot on the Choe data following RMA normalization. Two interpretations of each of the conditions of the Golden Spike are illustrated here. The "treatment" and "control" data were generated by normalizing both conditions (6 chips) simultaneously, and then plotting each condition individually. To produce the "control alone" and "treatment alone" lines, RMA was performed on each of the conditions individually (3 chips). Although this is less than the recommended minimum for an RMA procedure, we were interested to see whether the different variance vs intensity relationships were produced as an artifact during normalization. This does not appear to be the case.