That's how I used to do it when I used Audition.
Note if you're also keeping 24-bit files for playback, it may not make sense to dither (though it probably doesn't matter a lot, and most of us probably wouldn't hear the difference). Audition only converts to 32bfp those portions of the file on which it has performed an editing operation of some kind. So if I only applied fades in/out, for example, I just truncated from 32bfp to 24-bit. 99.9% of the file - everything except the portions I faded in/out - remained at 24-bit, since I didn't do any editing, so no need to dither needlessly. (To confirm this, edit a small portion of the file. Then select <1> the edited portion of the file, and <2> a non-edited portion of the file, and run Analyze | Statistics - you'll see <1> shows 32bfp, and <2> shows 24-bit.) If I did any substantial editing, I would apply dither normally, as you noted.
And if you're interested in other dither options, here's a comp:
http://taperssection.com/index.php/topic,51476.0.html