I got to poking around this issue today. Sadly, the graphs that used to be available at some M3 site seem to be NLA because the site is gone. I did find a reference tho to rear camber gain being 0.9deg per inch of compression. The problem with that # is that it came from a guy with sport springs and we have more drop then he does. This is an issue the camber gain is almost sure to be non-linear, therefore his 0.9deg per inch of compression might not be accurate for us. My recollection of those graphs is that they also showed that camber gain was nonlinear.
With my car on the lift I raised the car up in ~10mm increments and measure camber each time. I also put a 100lbs of weight on a rear corner and measured the camber with the 4mm of compression that resulted.
Bottom line. The 0.9deg of camber gain per inch is a reasonable ballpark. Maybe for an inch or two of deflection the relationship is close enough to linear for government work. If the nonlinearity of the camber gain makes the wheel gain less or more camber in it’s last inch of travel, my experiment didn’t conclusively show that. The #'s I got seem to indicate that we gain more in compression then in decompression, but standing here thinking about how a semi-trailing arm works, my feel is that it should be the opposite. So I’ll split the difference and say that for all practical purposes, the camber gain seems to be linear.
I looked at a rear spring for a moment and guessed that it would be willing to compress about 2.5" before the coils bound up. Call it ~2.3deg of camber gain at full compression.