Hi,
my advice is: if you want to get acceptably accurate results, don't do a tensile test in order to determine the rupture force (in shear) of your pin.
Rather, reproduce with a suitable machine the real fashion in which the shear-pin is acted.
Moreover, don't test a single specimen: as it has been said, shear-pins are sensitive to the notch diameter, which can vary because of tolerances, to the surface finish (more constant, but who knows...) and to the material characteristics of course (which can vary slightly even in the same bar...).
Where I work we use extensively shear-pins as security devices, and the procedure we follow is like I've said: 5 - 10 specimen from the same bar from which the definitive ones will be taken (this makes a 1-to-4 / 1-to-8 sampling), shear-machine applying the force in the exact same way as the "real" system. Nevertheless, we have frequently 5 to 15% dispersion in the results...
Regards