To be more precise: > Bin-by-bin correlations are not taken into account. correlations are not taken into account depending on the S/P ratios you provide. More specifically, if you use what comes out directly from the S and P fits in data/MC enriched/depleted samples. In this case, correlations are not taken into account. If you fit the mx distribution of the correction factor with n-th order polynomial, you can take correlations into account. I have done this by using the macro fitDRplot.C in VirVubFitter/fittest. The macro lets you choose the order of the fitting polynomial (from 0th to 3rd order) and gives you a file with the fitted S/P numbers as a function of mX, where correlations between fitted parameters are taken into account, and which can be used as input in VVF. Cheers, Concezio.