Yesterday, I speculated that maybe the difference between the pi0-triggered Pythia8 corrections and the pi0-triggered Pythia6 corrections might be a question of tracking resolution.  More specifically, that I'm overestimating the resolution in the Pythia8 case which makes the corrected data harder.  Details can be found here:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-07232019-run-9-pp-comparing-jet-response-between-pythia6-and-pythia8-pi0-triggers

To assess this, I turned off the pT smearing in our Pythia8 simulation, calculated a new response matrix and jet-matching efficiency using the "unsmeared" Pythia8 simulation, and then used this new response to unfold our pi0-triggered data.  The three plots below compare this new unsmeared Pythia8 response to the pi0-triggered Pythia6 response.

And this plot compares the data unfolded with the unsmeared Pythia8 response to the data unfolded with the pi0-triggered response.

The agreement between the two distributions is at roughly the same level as when I used the "smeared" Pythia8 response and the hadron-triggered Pythia6 response.  However, the agreement at the data point is definitely improved, but this is probably due to using the pi0-triggered Pythia6 response rather than using the unsmeared Pythia8 response.  The original comparison (between the smeared Pythia8 response and the hadron-triggered Pythia6 response) can be found here:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-07102019-run-9-pp-comparing-data-unfolded-pythia8-priors-vs-pythia6-priors

This rules out the effects of smearing.  So the next step is to take an especially critical eye towards the jet-matching algorithm...