During rearing of broiler breeders often a form of grading is applied, separating the smaller and/or bigger birds from the group to allow them to catch up/slow down and in that way improve the uniformity of the flock. There is a lot to say about the grading procedures, when to do, how often, how to adjust the feed for the different groups etc, and on this website several publications can be found that are dealing with these aspects.
However, we should realize that what we actually do with grading is that we separate the birds to reduce the competition in the flock. All birds in a flock have a more or less equal genetic background and therefore a more or less equal chance to become a normal, average bird. If a bird does not eat its fair share as for some reason it is not as competitive as other birds, the body weight will remain below the average. To allow that bird to catch up again we separate it into a group of equal small birds, and in that way reduce the competition and give it a chance to eat the amount it is supposed to get.
This results in a group of equally small birds that face less competition from the bigger, more agressive birds and are therefore more likely able to eat what they should. At the same time, the birds that remain in the normal group are now getting more competition as the smaller birds are taken out, and we normally see that the real big birds in the group are not getting that big anymore. They face more competition than before as the real small birds have been taken out, and their feed quantity available is less as the non-competitive birds that were eating less than the average are not there anymore, leaving less feed available for the bigger birds.
It is helpful to realize that grading is actually about reducing the competition, as it has some consequences. When we have very big populations of birds, it seems that the competition is stronger than when we have relatively smaller groups. This means that making big groups of selected birds is not as effective as making relatively small groups, as in the big groups a new level of competition will take place. It also means that when we have bigger groups of rearing birds, it is probably more effective to grade into 3 groups (small, average and large). When we have relatively small houses a grading into small and average might be sufficient. It is difficult to give exact numbers for it, but in my experience a group of small birds should not exceed 1000-1500 individuals, to avoid that new competition will take place. A group of average birds can be bigger but should not exceed 6000-7000 birds.
This holds for females, but the situation in males is slightly different. Males are much more agressive and competitive than females, and the chance that in a bigger group there are males that cannot compete is therefore bigger as well. We solve this by reducing the stocking density, but we also have to consider our grading procedures. Where in females selecting 20-30% of the smaller birds is often sufficient to reduce the competition and have both groups (average and small) relatively stable, this seems to be too little in males. We usually see that in males we get better results if we grade in two equal groups, smaller and bigger in 50/50 quantity, so that also in the bigger group the competition remains better under control. When we grade males in the smaller 25% and the bigger 75% of the population, we often see that in the 75% population we again get a shift in smaller and bigger birds due to competition, and we have to grade again at a later age.