Thanks for the question, Chris. The 75% figure is related to bound stellar systems (aka clusters). It is a
log-normal distribution function (sorry about that) determined by he properties of star clusters as bound systems versus associations as non-bound. In a bound system the more massive stars clump toward the centre in part because they are more massive, but more important because they drive away the dense formation gas that made them in the first place. The mass of the gas ejected can be two to three times the mass of the remaining stars. This makes for a significant mass-energy imbalance. The technical term is
virial disequilibrium. Several things happen as a result: (a) orbital densities become tight in the central region and many of the massive stars form close binaries whose orbits are only a few tens of their diameters, hence they rapidly rotate, (b) close binaries bulge into teardrop shapes facing each other called Roche lobes, which fill with envelope gas, (c) the quantity of this Roche gas varies depending on the stars' distances, but generally is 1/4 to 1/3 the total envelope surface mass; (d) the Roche gas is thinner and cooler so it no longer radiates as much light per unit surface area, and (whew) finally, (e) the log-normal distribution for the combined number of stars versus the total amout of emitted light comes out to the peak of a curve at 75%. It's a calculated rule of thumb that has been shown to have utility. Other calculated functions have found utility as well, e.g., the half-light and half-mass radii, the dynamic time (orbit time of a typical star) and crossing time (time it takes for a typical star to cross the cluster). The word 'typical' in this context is a little like faces in a crowd: each is different close up, but they mostly look the same at a distance.