Researchers use stratified random sampling to obtain a sample population that best represents the entire population being studied. Stratified random sampling involves first dividing a population into subpopulations and then applying random sampling methods to each subpopulation to form a test group.
Stratified random sampling is different than simple random sampling which involves the random selection of data from the entire population so each possible sample is equally likely to occur. In contrast, stratified random sampling divides the population into smaller groups, or strata, based on shared characteristics. A random sample is taken from each stratum in direct proportion to the size of the stratum compared to the population.
Example: Researchers are performing a study designed to evaluate the political leanings of economics students at a major university. The researchers want to ensure the random sample best approximates the student population including gender, undergraduate and graduate. The total population in the study is 1,000 students and from there, subgroups are created as shown below.
Total population = 1,000
Researchers would assign every economics student at the university to one of four subpopulations: male undergraduate, female undergraduate, male graduate and female graduate. Researchers would next count how many students from each subgroup make up the total population of 1,000 students. From there, researchers calculate each subgroup's percentage representation of the total population.
- Male undergraduates = 450 students (out of 100) or 45% of the population
- Female undergraduates = 200 students or 20%
- Male graduates = 200 students or 20%
- Female graduates = 150 students or 15%
Random sampling is done for each subpopulation based on its representation within the population as a whole. Since male undergraduates are 45% of the population, 45 male undergraduates are randomly chosen out of that subgroup. Because male graduates make up only 20% of the population, 20 are selected for the sample and so on.
Stratified random sampling accurately reflects the population being studied because researchers are stratifying the entire population before applying random sampling methods. In short, it ensures each subgroup within the population receives proper representation within the sample. As a result, stratified random sampling provides better coverage of the population since the researchers have control over the subgroups to ensure all of them are represented in the sampling.
With simple random sampling, there isn't any guarantee that any particular subgroup or type of person gets chosen. In our earlier example of the university students, using simple random sampling to procure a sample of 100 from the population might result in the selection of only 25 male undergraduates or only 25% of the total population. Also, 35 female graduates might be selected (35% of the population) resulting in underrepresentation for male undergraduates and overrepresentation for female graduates. Any errors in the representation of the population have the potential to diminish the accuracy of the study.
Unfortunately, stratified random sampling cannot be used in every study. The method's disadvantage is that several conditions must be met for it to be used properly. Researchers must identify every member of a population being studied and classify each of them into one, and only one, subpopulation. As a result, stratified random sampling is disadvantageous when researchers can't confidently classify every member of the population into a subgroup. Also, finding an exhaustive and definitive list of an entire population can be challenging.
Overlapping can be an issue if there are subjects that fall into multiple subgroups. When the random sampling is performed, those that are in multiple subgroups are more likely to be chosen and as a result, would be a misrepresentation or inaccurate reflection of the population.
The above example makes it easy; undergraduate, graduate, male, and female are clearly defined groups. In other situations, however, it might be far more difficult. Imagine bringing characteristics such as race, ethnicity or religion into play. The sorting process becomes more difficult, rendering stratified random sampling an ineffective and less than ideal method.
For more on random sampling, please read "The Difference Between Stratified and Simple Random Sampling" and "Examples of Stratified Random Sample."