Thanks for making the effort to test it in this way.
Because that design turns it more into a cantilever beam and assuming the force at the center of the bar is the same in all cases, simply moving it closer to the tube by mounting the bushing vertically as seen in the RMS Alter-K-Tion is an improvement (though the sway bar is very low hanging otherwise), it's just flat out deflecting less. Let's say that with weight transfer included the car might have 800 lbs by jacking up one corner in a static condition.
View attachment 1716197929
The issue with the parallelagraming is -> that is the most weak direction of the car and where its quite relevant for cornering, if, for example, there was no k-frame, the car in front of the windshield is kind of like a cardboard box with the top and the bottom removed. It would parallelagram or even splay very easily. The K-frame of any type helps resist it, and the k-shape would be harder to twist with a load applied at the LCA pivot point than having a simple crossmember at a right angle. The more the load is behind the rear most bolt, more lever arm, meaning more flex each time. The stock one the arm is behind in the same way, but the pins are directly behind the rear bolts for the k-frame. It's actively reducing the twisting directly into both the k-frame and the chassis without any cantilever lever arm. This is where adding something behind comes in because it would generally stop any of this twisting action relative to the body and spread out the load. You can see this with something like the MM mustang K-member. It's not possible with the OE type oil pans and the available headers to have the crossmember go directly through the LCAs which would be an alternate design.
I don't know the reason for the cross bar having a rod end. The left to right side can twist meaning if, for example, you went into a corner and the left side mount is attempting to move in an arc due to the twisting of the k-frame about the axis going front-rear in the car Since there is a ball there, it just keeps the straight line distance to the other side the same only in the cross car direction, it will do zero for any simultaneous front-back movement because it will just pivot on the rod end instead of having to bend the rod to deflect. With only one bolt on each side, it can also run an arc. The degrees of freedom are not adequately restricted to get the best possible effect. For example, my truck has a removable crossmember similar to the transmission crossmember on our cars in roughly the same area and it has two bolts on each side with what's essentially a box tube in the middle. It has no degrees of freedom other than literally twisting the steel.
The frame rails themselves are pretty weak and nothing like say a hydroformed pickup truck frame. They're just supported by the inner fenders. The cars flex all over the place which is why we have seen cars with high wheel rates cracking the spot welds or inner fenders. It just fatigues out with the higher loads.
