Hi Jacob
I think you were right first time: complex 'structure factor' = real 'structure amplitude' with real phase. You are presumably not interested in the distribution of structure factors, only the amplitudes, since the former would also depend on the phase distribution. As pointed out by others, the form of the I and |F| distributions for acentrics derives from the fact that the acentric structure factor is a complex number with 2 degrees of freedom. So your original statement concerning amplitudes is correct, with the exception that your statement that the |F| distribution is normal is only true for centrics, not acentrics or twinned reflexions.