It's a relative measurement. The new work is 3dB lower or higher than the old (for example). From my reading of your post, there's still no good way to tell from what.Is this not an accurate measure of SNR?
Without that, nobody can replicate your results.
Koz