Hi Folks,
I agree that it's appropriate to quantify the improvement in noise reduction. It's also hard to do: the reduction will depend on frequency, and on making a precise duplication of ambient noise to benchmark the reduction. My original intention was to drive the car in multiple opposing passes on the same stretch of moderately bumpy road, and at the same speed (noise almost invariably increases with increasing speed,) to standardize the excitation. My opinion now is to use a speaker instead to reproduce pink noise, placed at some fixed relative position to the car (most likely about 6' away from the side of the driver's side of the car, on the car's transverse axis,) in a quiet environment where the car is likewise fixed with respect to the environment. The measurement would be before/after spectral response with 1/3-octave smoothing using an ANSI-standard spectrum analyzer, at the average location of my ears, in the driver's seat. I'll probably use either no weighting or C-weighting, so I can also report single numbers for before/after SPL.
Short of this, sad to say, most of the subjective opinions being posted here are not of high confidence. That's not to impugn anyone's opinion, but it is an acknowledgement of the fact that acoustical perception is notoriously unreliable.
I already have the very high loss-factor butyl rubber sheet that I intend to apply to the entire inner door skin as the noise barrier. Judging by the sheet's very high creep and extreme deadness, I believe, but am not sure, that it is IIR, my favorite for this application. The only thing delaying me here is admittedly laziness: I've had this sheet for a long time, but haven't felt motivated enough to actually follow-through on this plan. Hopefully some self-imposed posting here will help me got off my duff...
Thanks,
Mark