It is usually very improper to just pick one number indiscriminately, in light of the fact that they aren't all similarly likely. So it's good to have a good and size number of samples in order for the data to not have too much errors. And doing this will help arrive at a better conclusion for the study that is being conducted