No, that's quite correct. There's a huge difference between the format that the mixer works in internally, and what bit depth a file is saved as. The only reason the mixing bit depth is listed at all is historical - Audition used to be capable of doing 16-bit integer mixes, but there was so much confusion caused when people did this and ended up wrecking files that a decision was made a long time back now to do all mixes internally as 32-bit Floating Point ones. These are essentially lossless when it comes to level changes, which integer ones most certainly aren't - you can lose a lot of bits very quickly with integer mixes! So essentially your final mixed output is streamed from the mixer as a 32-bit file, but there's nothing to stop you saving it in any format you like or need. Ideally it's good practice to store at least a copy of it as a 32-bit file, even if you don't necessarily need it at the time - that's the highest quality and most flexible file you'll have. If you are desperate, I can tell you why, but it's technical and essentially boring...