Sampling, for the purposes of this guide, refers to any process by which members of a population are selected to participate in research. There are many methods for sampling, each with a slightly different purpose. In the box below you can learn more about these common sampling techniques:
Before you can obtain a sample, you must first identify a target population. The target population refers to all of the people who are the focus of a study. For example, a study about elementary school teacher burnout would include all elementary school teachers in the population. In some cases, you may need to consider an accessible population. This is a subset of the target population that can reasonably be accessed by the researcher for sampling. Oftentimes, researchers will use a sampling frame to facilitate their sampling methods. A sampling frame is a list of all of the members of the population.
Through simple random sampling (SRS), all members of the population have an equal chance of being selected. Therefore, this is a type of probability sampling. A rudimentary method of SRS is drawing names out of a hat. Each slip of paper has the same chance of being chosen on every draw. You could also use a random number generator to facilitate random selection from the population.
Simple random sampling assumes that all members of the population are accessible. If your population is "people in the United States" and you are attempting to sample via the Internet, members of the population without Internet access do not have a chance to be selected. This would not be an appropriate use of simple random sampling.
Researchers use SRS when the intention is to obtain a representative sample that can provide data for generalizing to the population. If members are chosen randomly, the sample is less subject to bias that may exist by non-random sampling methods.
Stratified sampling is a two-step sampling procedure. First, the population is divided into groups or strata. How this is done will depend on your specific population. Using the example of elementary school teachers, we could divide the teachers up based on state or school district, with each state (or school district) representing one strata. Next, members of each strata are selected for participation. When they are selected randomly from within each strata, it is called stratified random sampling.
In the above figure, the population was divided into four strata. Members of each strata were selected to participate using simple random sampling. Thus all four strata are represented in the final sample of participants. This method is effective in ensuring all strata are included in the sample. For example, making sure teachers from all 50 states are included in the sample.
You can also adjust the proportion of the sample that comes from each strata to maintain proportional alignment with the population. For example, if 50% of the population is in Group Three, 50% of the sample can be randomly selected from that strata, thus ensuring the final makeup of the sample aligns with the makeup of the population.
Like stratified sampling, cluster sampling is a two-step sampling procedure that also starts with dividing the population into groups called clusters. As with stratified sampling, how groups are divided will depend on your population. For example, teachers could be divided into clusters based on school district or grade level taught. The primary difference between stratified sampling and cluster sampling is that whole clusters are randomly selected and everyone in that cluster is included in the sample.
In the above figure, the population was divided into four clusters. Two of these clusters were randomly chosen. All members of Group One and Group Four will participate in the study, making up the sample.
Cluster sampling is ideal when there are not major differences between the clusters. Consider dividing teachers based on school district versus grade level taught. Each school district will include the same grade levels, though may have some variability in factors like school size and geographic location. Selecting school districts at random can help create a representative sample that covers an array of factors. If the clusters are grade level, randomly selecting grade levels mean entire grades are not included in the sample. These grades may have meaningful differences from the included grades. So, cluster sampling may not be as effective in this situation.
Systematic sampling occurs when participants are selected at set intervals. For example, choosing every third person from a list. To ensure this method aligns with probability sampling conditions, the starting point is randomly selected. Consider the following visual that shows systematic selection, beginning with the second person in the line (the randomly selected starting point).
Systematic sampling offers benefits similar to simple random sampling but is often perceived as being simpler to carry out. It also combats the potential problem of clusters that can occur with random sampling. While random sampling aims to select a variety from the population, there is also no way to regulate who it selects. So, clusters of individuals could be selected at random, thus potentially biasing the research. Systematic sampling ensures and even distribution across the population.
