Additionally, you will notice that one of the shots has more white area than the other - the matrix (Canon = evaluative) meter responds differently to the amount of area. Considering that the matrix is also biased by the AF target (on the presumption that whatever is in focus is probably the most important part of the scene), you really have little idea of what the meter is doing without some experience.
You would have had similar trouble if the cars had been all black - although in that case the reverse would have happened, you'd have gotten overall overexposure, but for largely the same reason.
Note that polarizers have no effect on reflections off of metal. Whether or not the paint jobs count "as metal" may depend on how much metallic content they have. These don't appear to be metallic, so the painted surfaces may have been polarized, but of course the chrome isn't going to be affected.
> I'm a bit surprised and disappointed with P mode shooting and with the unintended consequences of AF-A focus. I chose those modes because of the automation they provide.
As you are discovering, AF and metering are both pretty complex things. As soon as you give up control to the automation, you have to understand what the automation is doing. For this reason I almost never use most of the more advanced modes on the camera, because I just don't have the experience to know what they're going to do with any precision. If I really need to be sure of the result, I usually go back to spot metering, manual exposure, and at most single-point AF.
> I'm not in the habit of scrutinizing the viewfinder for camera data. Just for framing.
My advice is to start getting into that habit. Knowing what the camera is doing or about to do is very valuable info. It just takes some time and effort - and that starts with knowing why it's important.
_____ Brian... a bicoastal Nikonian and Team Member
My gallery is online. Comments and critique welcomed any time!