I could even see 1% or so. But an almost 10% loss just makes no sense. That's like putting in subframe connectors and losing rigidity.
I understand your skepticism (and share it, to an extent) but it seems the ball is in the doubter's court- either find fault in the methodology or replicate the experiment with different results.
Keep in mind also that the Z chassis is not directly comparable to the much later FC/FD units.
First of all there's no b-pillar (which is why nearly every Z you see will have a crack at the roof> rear quarter joint) and the chassis lacks the deeply drawn structure that modern metallurgy and forming allows.
It
looks like a modern unibody but lacks the benefit of structural analysis/development available today.
*Bear with me, the following is an analogy...*When radials began to supplant bias ply tires, British sports car owners started popping their 48 spoke Dunlop wire wheels...the extra grip and braking force just overwhelmed them.
It's possible that seam welding eliminated enough body slop to concentrate the forces in a smaller, already vulnerable area, which now deflects more easily than before.
Just a guess, based on the assumption that the data as presented is correct.