General rule of thumb: under white light, plastics and other non-metallic surfaces tend to have white specular highlights, whereas metals tend to have specular highlights which are a similar shade as the metal. (Or to be more pedantic, most of the color of a metal is actually its specular reflection; they tend to be quite dark, in terms of their diffuse color)
Once paint is on the metal, though, all bets are off.
For example, opaque paint might reflect light exactly like plastic does, even though there's metal underneath. If the light's not making it down to the metal, then it doesn't really matter at all that there's metal down there.
Similarly, it's very common to put a clear coat of paint on top of metal ("clearcoat"), which will add a white specular highlight, similar to the one you get on plastic, with the metallic light behaviour underneath, on the next layer down. (Disney is fairly well known for putting simulated clearcoat on virtually every surface in their recent 3D-rendered films)
Car paints are often translucent; some light will interact with the paint, some will pass through to the metal underneath, and the light which passes through to the metal underneath and is reflected will pass back through the paint again on the way back, being affected in both directions.
Phong was one of the first algorithms for estimating specular lighting, and it's far from "state of the art", these days; it can't model any of these more complicated surfaces to any degree of accuracy. But if you're not after accuracy, it can still do stylised lighting okay, as long as you control the size and color of the highlights to be appropriate for the materials you're simulating.
As a rough rule of thumb, in the absence of other factors, and under a white light, plastics tend to have a white specular highlight, while metals tend to have a colored one.
