Let say if we killed about 5/6 of the population. How would the results differ if we only killed ~1/2 of the population? If the phenotype frequencies in your survivors was very different than the phenotype frequency in the starting population would you expect this difference to be less or more if you killed less individuals (had more survivors).