Collins Systematic Leak Through

To calculate the amount of leak through error due to a Sivers asymmetry in the Collins analysis over the Run6 transverse data, I'm trying to implement a method used by Jim Drachenberg outlined in his thesis paper. This method entails creating a sample of "generated" yields of pions in jets. These are created from the true physics yields, but filled with both spin states. For a generated yield we will fill it with the true aligned spin state and weight it with (1.0 + A_siv) where A_siv is the true measured, unpolarized Sivers asymmetry, and then we also fill it with the true oppositely aligned state and weight it by (1.0 - A_siv). This should effectively create an unpolarized envrionment, from which we can form a cross-ratio from the generated yields. 

To ensure that the method is working appropriately we can bin the cross ratio in phi_S instead of the Collins angle. This, in turn, should reproduce our input A_siv asymmetry when it is fit with a polynomial -- p0 + p1*sin(phi_S). Attached are plots that check that the input is what we extract when binning in phi_C. 

The two plots of input vs. extracted are perfectly linear as we would expect when we have a 1-to-1 correspondance between consecutive input and extracted values. Also, the histograms show clearly the fits used and the extracted parameter.
Looking forward we would like to bin this in phi_C and see how the asymmetry values come out when extracted. We would expect with no Sivers contamination, that we would return a value of zero, but any asymmetry extracted will be filtered into the systematic error.