The reason these cluster methods work is that we make the assumption that star clusters are physically associated groups, and all cluster members are approximately the same distance from us. If the clusters were just chance visual groupings then these techniques would fail. Of course, every time you look at a cluster you see it, plus all the stars between us and the cluster, plus all the stars that lie in the line-of-sight behind it. To do meaningful statistics, you have to identify and discard these guys and not include them in your calculations. There are techniques to do this too, but none are foolproof. Astronomers are very clever people, and they earn their money.