I suppose you could argue that an arbitrary value is chosen for the g forces encountered in a "standard crash" so the process is inexact on that basis, and so some variation in pax weight doesn't create a step change in the validity of the certification.