I have a hypothesis, based on a single incident, and difference in alloy, as mentioned, seemed to make a difference, BUT it seemed to mostly be the die.
I opened a Lee .430" die to .4325", which seemed to work just fine on softer alloys, like WW/Pb. When sizing straight WW or harder, they came out .433" - .0005" larger than when using the softer alloy. Since I took it for granted that the bullets were coming out .4325", I did not measure as I sized. LATER, I wondered if they had "rebounded." Double/triple checked mics and zeroes, wasn't the measuring tools.
What I THINK happened, is that when I opened the die with abrasive paper and a wooden dowel, I created a shorter, rounded minimum diameter in the die, instead of a longer, straight section, like in a Lyman die. What I believe happens is that the shorter section of minimum ID displaces metal ahead of and behind the portion being squeezed, such that the OD of the bullet is not "springing back," rather it is displaced forward of the minimum ID as the bullet passes through that choke-point. This happens with the harder stuff, not the softer stuff. I have been careful since to use a metal rod, as large as possible, to hone or lap dies to larger IDs in an attempt to maintain a longer section fo minimum ID.
Basically a somewhat informed conjecture on my part. Not sure it exactly addresses the actual question, but may contain a clue for someone who possibly has another puzzle piece it fits into.