does anyone know the best way to test handling?

as in cutting out driver error as much as possible from test to test?

changing out the sub-frame i think it is, to a traction bar, and the company is saying it will improve handling, is there any objective way of testing this with somewhat hard data?