In the test I refer to, which was a large fabricated panel with several stiffeners under compression, whole joints unzipped. There was an equal mix of pull-through and shank failure. It was not really possible to identify precisely where it began - i.e. whether the initial failure was pull-through or shank tension. Some realistic measured allowables, with their failure modes, is really what I am looking for. It seems that the failure was premature with respect to, for example, the NASA manual values.