Fig. 1. Influence of propofol and sevoflurane on Observer's Assessment of Alertness/Sedation Scale (OAA/S) as Trellis plots. 54Raw data from nine subjects are shown in boxes as circles. On the abscissa, the concentration of the administered drug is given as a fraction of the C^{50 awake}concentration: MAC^awakevalue of sevoflurane (lower row), CP^{50 awake}value of propofol (upper row). Note that for propofol, calculated concentrations were used. Data in corresponding boxes in the upper and lower row are from the same subject. Missing data are handled correctly by the mixed-model ANOVA, which still works, although sevoflurane data were not available in the second subject (blank box), and propofol was not administered to the sixth subject. In the model used, the slope is set as a fixed parameter for each drug, i.e. , regression lines have the same slope for all individuals. For each effect, the drug with the steeper slope would be the more potent drug. In this plot of OAA/S, the slope of sevoflurane is significant (as tabulated in the second column in table 1). The difference between the slopes of the two drugs (as tabulated in the third column in table 1) is almost zero, indicating that both drugs have similar potencies. From the definition of the OAA/S and the normalization of concentrations to C^{50 awake}, we expect a slope of −2.5 (regression:−3.01) and an intercept of 5 (regression: 5.02) for both drugs, which agrees well with the values from the regression analysis and demonstrates that the C^{50 awake}normalization we use is consistent with the data of the subjects in our study.