Sometimes it is not possible for a researcher to pool all possible subjects into a single list and sample subjects from that list. In many instances of real-world research, subjects are naturally grouped into areas or organizations. In these cases, staged sampling proves much more convenient than other forms of sampling. For example, instead of sampling at random from the population "the trees of the Pacific Northwest," it is much more convenient to sample a few areas in the Pacific Northwest, and then randomly sample trees from those specific areas.
In many studies, the most expensive and time-consuming part of the study is the sampling procedures. Any methods that can cut down on cost would be advantageous to the researchers. Staged sampling is such a method in cases where the population is large, and proves to be a reliable alternative to cluster sampling. In fact, staged sampling has the advantages of cluster sampling (in that it divides the population into groups), but is more efficient in terms of time and money due to it avoiding the sampling of all individuals in the determined groups. In this way, the method of staged sampling cuts back on interview time and compensation for subjects.
Staged sampling, because it divides the population of interest into distinct groups, lacks representation power. The problem lies in that when staged sampling takes place, specific groups are unable to enter the sample. This means that information about certain demographics may be lost. For this reason, researchers avoid staged sampling unless they are certain that within-group variance is higher than between-group variance, as otherwise results would not generalize to the main population.
Unlike random sampling, where all decisions as to what individuals to include in the study are made through random processes, staged sampling requires the user to make the decisions. This is problematic for two reasons. First, by making decisions, the researcher has introduced subjectivity into the study, which may bias the results. And second, there are no real procedures for choosing the "right" groups, sizes of groups or subgroups. Thus, researchers have difficulty justifying the decisions made in the process of staged sampling.