Response matrices extracted from the run 12 dijet embedding sample.  These plot the raw jet pT of pi0-triggered full recoil jets for R = 0.4: the x-axis is the detector level pT, and the y-axis is the corresponding particle level pT.  The various panels show the response matrix for various hadronic correction schemes.  Ideally, we would like to use the %-correction which produces the tightest correlation between detector and particle level jets.  Judging by the plots below, it would seem that the 100% correction scheme satisfies this as it has the smallest RMS along the x-axis...

Note: Response matrices are normalized per trigger (i.e. each bin of particle pT is not normalized to one).