I get the positive feedback loop issue. I’m saying if I’m looking at the data (which is blind a -it doesn’t classify people by ethnicity) but manually when looking at and labeling the data we find the above proportions as the effective proportions, then I think we’re okay.