Same as Fig. 7 but with the SFR determined from the UV (observed, so attenuated) and IR emission, following Kennicutt (1998) standard conversions between L_UV and L_IR and SFR. This mimics what is being doing when computing the SFR from the UV and IR emission of galaxies. Taking the “observed” SFR (this figure) rather than the “true” SFR from the model (Fig. 7) decreases the dispersion and removes a large fraction of the outliers.