When you combine:
• clipping
• possibly limiting
• possibly aliasing
• bandwidth limiting
You get such a complex combination I wager no one can predict the results. Many years ago I moved over to frequently using analog non-linear processing over digital non-linear processing because I felt the sound via analog was cleaner and more open, and that the analog processing seemed to tolerate purposeful or accidental clipping much more nicely. Reasons why include: lack of aliasing, softer action (lower effective ratios, longer attack times), personal listening preference.
I soon learned, however, that pushing the ADC above 0 dBFS — even a few tenths of a dB — was a crap shoot— if the music was complex it could mask the distortion, but single sounds (flute, piano, solo snare) quickly sounded ugly. So I moved to either avoiding clipping or to using analog peak limiting instead.