If it is avoidance of inter-driver modulation and maximisation of star-earthiness, then the results will depend on the existing 0V arrangements. For my E-IIIR, each driver's 0V is connected to a common 0V point on the XO board, which is then taken connected to the amp's negative terminal. For this to be improved upon, I need to run the 0V returns of each driver directly to the amp. Hence, my suggestion in post #35. So a very short run of very low DCR cable between the amp and XO could be a valid proxy for direct grounding.
That would mean three separate runs of cables from the XO enclosure to the loudspeakers in my case, whereas I had conceived using a single run of four-core cable (or six-core, doubling up conductors for 0V and + for LF), but I can't see how that improves my current set-up, which has a 7m run of 4-core #15AWG Mogami W2972. I've doubled up the conductors, so they are closer to #12AWG with DCR of 0.0044Ω/m. Over a 7m run, that's a barely measureable 0.03Ω.
It would be useful to know the LCR characteristics of cables used by those who reported a difference.