For the question below disregard the quality of sample rate conversion and bit reduction by the DAW (Live 7) versus r8brain.

Recording in 24 88200.  Then when ready to go to 16 44100 using Elephant's dither in the final track and recording that track to a new track using the DAW to do the 16 44100 conversions.  I am assuming the DAW will first be processing the 24 88200 track with the Elephant dither, then resampling and the converting to 16 bits.  Is there anything inherently flawed in doing it this way?  Would it make sense to convert to 32 88200 in the DAW first and then go to 16 44100?  As far as convenience it beats using a non DAW dither (eg r8brain), and then going back into a new session of the DAW to use it.  I admit having a hard time noticing sound differences in different ways of going about it.
Rick

Dithering should always be applied last - before bit truncation occurs.  If you do resampling before bit truncation, dithering should be placed AFTER resampling - not before it.

Thanks Aleksey.  I guess it's two steps then, 88200 to 44100 and then Elephant while going to 16 bit.
Rick

Yeah, sounds fine.