If I understand correctly, you were recording to DAT at 16-bit / 48 kHz and switched mid-recording to (let's assume for now) 16-bit / 32 kHz.
I'd split the file at the moment you switch from 48k to 32k, process each file separately (downsampling 48k to 44.1k, upsampling 32k to 44.1k via Convert Sample Type), then re-join the files.
If you're -not- going to do any additional edits, it's okay to perform the Convert Sample Type function at 16-bit resolution (Audition will perform the operation at 32bfp and then internally dither down to 16-bit automatically).
If you -are- going to do any additional edits, you should perform the Convert Sample Type function on each file at 32bfp, complete any additional editing, and then finally dither down to 16-bit.