What exactly are the other differences / issues between the original and corrected measurements? If all these issues were present with all previous speaker measurements, that may give hope of retroactively correcting them. By subtracting the original Genelec data point SPL values from the corrected ones, you would arrive at a correction curve for the original measurements. You can then do the same for the next (few) speakers by measuring them using both the original and corrected method, to find the correction curves for them as well. If all the correction curves are the same (or similar enough within the fitting error of the measurements), then I think you can safely assume this correction curve is not speaker dependent, and so it can be retroactively applied to all previous measurements, in order to allow for an accurate comparison with the Genelec and all future measurements, and to produce fair, reliable preference ratings across all speakers. By the way, thanks a lot for identifying and attempting to rectify this issue (and thanks to the vigilant forum members who originally spotted the systematic errors in the measurements).