Using the microwave oven test is nice to compare two or more materials, but you still don't know whether the best (or worse) material matches your requirements. When the shape of the materials differs, you may get strange (unreliable) results.
Properties of paint amaze me also (huge difference in reflectivity and transmissivity). Some (spray) paints with metallic look do virtually not block the (UHF) signal.
**They use flakes of stuff with optical properties like mica, or "fools gold".
I ones used black thick polycarbonate sheets as an antenna cover (mechanically heavy loaded). It did detune the antenna somewhat, but after retuning I couldn't measure the difference in produced field (w.r.t. uncovered situation, mid UHF situation). With black material I am always afraid to encounter too high carbon black levels.
**Black is the riskiest colour for RF welding of PVC. Clear (pure PVC is clear pale blue) is the most predictable. We assumed that the problem was the use of carbon black, but never found out for sure. Another problem is metallic inclusions, which resulted in spectacular fireworks and blew holes in the brass tooling.
**Handle also made a big difference. Rigid is hard to weld, and very soft is too easy and heat build-up in the tooling can melt the surface before the inside. Different fillers, such as chalk, make a difference, especially if they hold water.
**The chlorine is polar so the chains wriggle in an AC field. Shorter chains, especially when the structure is loosened by the addition of plasticiser, mean more wriggling. Presumably energy is lost in the breaking and remaking of whatever bonds hold the chains together.
**Dunno what this has to do with radio waves. It does illustrate how much variation there is in the material, as others have pointed out.
**The chlorine inhibits fire, incidentally. PVC in electrical installations can be fire-prone because of its electrical properties, but the risk of a serious fire is considered acceptably small.
Ian