So we now have access to a wide variety of gluon polarization parameterizations in our Pythia afterburner covering the regions between GRSV -std and -max and between GRSV -zero and -min.  Dave Staszak documents the details of the new distributions at

http://www.star.bnl.gov/protected/spin/staszak/2006/Sims/trigBias/trigBias.html

Here are plots of the raw charged pion asymmetries obtained from these grids (p070 excluded because of a bug on my part):

I suspect the problem with the pi-minus asymmetries can be traced to a lack of gg contributions.  For more information see Agreement Between "Polarized" Pythia and NLO pQCD.  In the remainder of this document I'll stick to analyzing pi-plus data.

I've attached multi-page PDF at the bottom of the page.  Each page of the PDF plots, for a give input gluon polarization distribution A_{LL}(trigger) - A_{LL}(no trigger) as a function of pion p_T.  There are six plots on each page; one for each trigger in (minbias, HT1, HT2, JP1, JP2) and then the raw A_{LL}(no trigger) distribution itself in the bottom right.  I believe one way we can obtain a systematic uncertainty from plots such as these is to calculate a double-sided max( delta A_{LL}, uncertainty on triggered A_{LL} ) for each pT bin we use to make our measurement.

One question is which grids are appropriate ones to use for calculating the systematic.  GRSV-max has been excluded in other measurements, so I don't think it's appropriate.  All other grids are fair game, which means the systematic will likely be derived from GRSV-min and p070 (once I rerun; use p060 for now).