Updated Plots and Tables

First, plots: Figure 1 shows my result, Figure 2 shows my result compared to Alan's (using the same pT bins), and Figure 3 shows the combined result (statistical errors only).

Figure 1:

 

Figure 2:

 

Figure 3:

 

I only showed statistical errors on Figure 3 because there was some question about how to combine the systematics.  Here's the breakdown of them, 2006 first:

Table 1: Systematic Errors for the 2006 Result.

Error Pt Bin 1 Pt Bin 2 Pt Bin 3 Pt Bin 4
Low Mass Background .001 .0011 .0038 .001
Combinatorial Background .001 .0086 .0016 .0003
Photon Energy Uncertainty .0015 .0034 .0017 .0015
Non-Longitudinal Components .0094 .0094 .0094 .0094

And now 2009:

Table 2: Systematic Errors for the 2009 Result.

Error Pt Bin 1 Pt Bin 2 Pt Bin 3 Pt Bin 4
Mass Window .0058 .0013 .012 .018
Relative Luminosity .0015 .0015 .0015 .0015
Trigger Bias .00028 .00059 .00097 .0017

 

Finally, there were some questions about how my result ended up so close to Alan's (the calculated chi-square value is .19).  Here are some tables of data, as requested, with 2006 first (note that I had to back the epsilon out of the final A_LL, the background fractions, and the background A_LL's: see here for A_LL and statistical errors):

Table 3: Information on the 2006 result

  Pt Bin 1 Pt Bin 2 Pt Bin 3 Pt Bin 4
++ Yield 20209 14088 7224 2505
+- Yield 20088 14057 7080 2470
-+ Yield 19961 13645 7107 2552
-- Yield 20231 13825 7204 2491
Combinatorial Background Fraction 5.8% 5.9% 5.3% 5.9%
Split-photon (Low-mass) Background Fraction 3.6% 3.9% 9.3% 8.6%
Raw Asymmetry .0065 .0007 .022 -.0065
Raw Asymmetry Statistical Error .01 .0125 .0164 .0268
A_LL .008 .0058 .0203 -.0084
A_LL Statistical Error .0115 .0136 .0189 .0306
A_LL Total Systematic Error .0023 .0038 .0043

.002

And the same for 2009:

Table 4: Information on the 2009 result

  Pt Bin 1 Pt Bin 2 Pt Bin 3 Pt Bin 4
++ Yield 42142 73951 57713 24689
+- Yield 41679 72765 55676 24856
-+ Yield 41509 72286 55967 24157
-- Yield 42049 73346 56401 25043
Combinatorial Background Fraction .064 .056 .116 .203
Split-Photon (Low-mass) Background Fraction .055 .078 .044 .072
Epsilon .0018 .0022 .0064 .00064
Epsilon Statistical Error .0037 .0029 .0036 .0056
Raw Asymmetry .0057 .0052 .017 -.00078
Raw Asymmetry Statistical Error .011 .0088 .011 .017
A_LL .0077 .0072 .023 .0037
A_LL Statistical Error .013 .01 .013 .024
A_LL Total Systematic Error .006 .002 .012 .018

Figure 4: Comparison of raw asymmetries (that is, asymmetries before accounting for background)

PS: If you compare my statistical uncertainties to Alan's, it seems like mine should be smaller, given the ratio between the yields I get and the yields Alan gets.  As far as I can tell, my statistical errors are larger than expected due to taking pion multiplicity into account (by filling histograms once per event, with the number of pions falling into a given pt bin for an event as the weight).  Unfortunately, I don't quite understand how Alan calculated his statistical uncertainties: it appears to be in code linked to here, but I don't understand why the formula given in this code is correct.  In his thesis, he says that he's accounting for the effect of pion multiplicity, and that it changes the statistical uncertainties by a few percent only: since he doesn't give a plot of the number of pions per event, I can't figure out whether this means that we're using different methods or if the multiplicities are genuinely different, although I don't think that my pion multiplicity should be so much larger than his, even with an expanded trigger mix.