Thermal degradation of polymers, if carried out far enough. is pretty straight forward – all the carbon is oxidized to CO2, the hydrogens are oxidized to H2O. As for the heteroatoms, oxygen is already mentioned, and nitrogen will go to NOx.
The more interesting results occur when the material is only partially degraded. I ran across a research highlight discussing the burnout of ethylene vinyl acetate (EVA) copolymer in powder metal molding. In powder metal molding, the metal powder is mixed with a polymer emulsion, molded and baked so that the polymer burns and the metal particle melt and flow together. Keeping the shape is critical during the baking step and the research found that certain polymers are better than others in this regards. The research looked at what specifically was occurring to the polymer during the burnout and found that the acetyl groups hanging off the polymer backbone broke off and (probably by abstracting an adjacent hydrogen atom) became acetic acid.  The result is a double bond forming along the backbone – what the authors call an ethylene acetylene copolymer. 
I was not aware of this chemistry prior to reading the writeup, but it actually isn't that new. Research has already been published on using the partial thermal degradation of as a means of preparing polyacetylene.
Given that the unsaturated polymer would be inherently stiffer than a saturated polymer (the geometric options around a double bond are far less than a single bond), it is not surprising that the partially finished metal objects are benefitting from this degradation product.
Lastly, lest you think that all substituted vinyl polymers degrade in this fashion, polyacrylonitrile follows a totally different route, one that is used to create carbon fiber (albeit, this is done in an inert atmosphere).
 Isn't this strange? Most polymer chemists are familiar with acetic acid forming during a polymerization step (of silicones most commonly), and not as a degradation product.
 A similar polymer forms when PVC partially degrades – HCl comes off leaving double bonds along the backbone. If there are enough double bonds that form alternating with single bonds, the resulting conjugation absorbs visible and UV light (depending on the conjugation length). Under the right conditions, the longer visible wavelengths are removed leaving only the shorter, redder ones. If the PVC is already white (such as is used in windows), you can end up with pink windows! (And unhappy customers!)