How I understand it is it's not 'movement' within the bushes but the unequal lengths of the suspension arms and therefore the changing geometry which causes the effect. The V6 has rose joints as they are harder on them due to increased performance and weight. Rose joints are stiffer than std bushes, as are poly bushes, but they don't reduce the effect so why should poly bushes?
Yes, I agree, it is the basic link geometry that determines the behaviour of the roadwheel as the suspension travels from extreme to extreme of movement - which is why, for accuracy and predictability, spherical 'Rose' type joints are usually only fitted on competition cars where this accuracy is important. I suspect that the GTV steer effect under discussion is roll angle induced onto the loaded wheel but need data to confirm.
Spherical bearings, as you know, are basically a steel ball in a nylon (GTV) or composition bearing material with no discernable play when new and they are capable of rotating through large , multi directional angular movements with the minimum of resistance. The link they are fitted to therefore remains a constant length under all conditions.
My point is this, the designer has decided what links need to be of this configuration to make the suspension work in the way he wants it to under all load conditions.
So, let's now substitute a rubber or 'poly' bush at either end. What happens? the link length varies slightly plus or minus according to the loading (which may be considerable with a full load and a 1G corner) and the bush resists rotation, less so radial to the axis of the securing bolt, and to an extent controlled by the stiffness of the bush material. This material has to allow sufficient link end movement relative to the bolt axis to accomodate angular variations during suspension travel but also offer minimum length variation on the suspension link centres.
Put flanges on the bush ends (top hat bushes)and another restriction to free movement is introduced (the 2Litre spring pan inner bonded rubber bush has no flanges and therefore does not introduce dynamic bending loads into the subframe lugs that they were not designed to take)
Add up all the little length variations due to load, resistance to angular change etc and you can see that the original suspension design becomes compromised in many ways - but of course it's tauter due to possible geometry conflicts and much quieter and the driver is pleased with the result as we see so many times in the forums!
It would be interesting to draw out the suspension geometry and analyse possible conflicts introduced by replacing these suspension components - but, hey guys, I can't be bothered to reinstall CAD on my computer as I'm happy with my spherical joints and sincerely congratulate the Fiat design team on a fabulous car that I feel very priviledged to own!!