The majority of the population is white, but that does not count out significant numbers of other races. Let’s pretend researchers want to conduct a study regarding race with results generalizable to the entire population of the United States. However, these researchers live in Maine, where 94.4% of the state is white. If they just used random sampling, all (or almost all) of their participants would be white. This study would be completely useless in generalizing to the broader American population. These researchers would have to use stratified sampling to ensure that the demographic of their participants matched that of the United States. They would have to search the remaining 5.6% of the state to find strata of other races to pull from. Once they found the few non-white people, they would then randomly sample from that group.

Hopefully, this situation would not take place in Maine, but it gives us a better understanding of the importance of stratified sampling. Without it, results from a study like this would be unable to be generalized past white people, even though the researchers did use random sampling to obtain their participants. Researchers could also conduct a study where each race is equally represented, therefore (intentionally) not reflecting the demographics of the United States, but hopefully, they don’t conduct that in Maine.

Steps for Stratified Sampling

There are suggested steps to follow in order to complete a study using stratified sampling.

1. Define your population: Who are you studying? The entire US population? College students? Football players?

2. Define your strata: What subsets are you separating your population into? Into religions? Socioeconomic status? Gender?

3. Separate the population into their specific stratum: Who goes into each stratum? Remember, each person can only go into one stratum!

4. List the sample size: How many people are you actually going to research from your overall population?

5. Determine necessary strata sample sizes: Do you want them to reflect the actual numbers of the population or do you want all strata equally represented?

6. Random sample within the strata: Take your samples from the strata to get your subjects for the study.

Stratified Sampling Example

Let’s pretend you’re in college and your psychology professor wants you to help her conduct a study on how people respond to a confusing situation with an emphasis on the overall results as well as how individual majors respond differently. You have all the details ironed out except the subjects of your study. You decide that you want the percentages of different majors partaking in the study to reflect the percentages of majors on campus.

Let’s simplify all the options of majors and pretend that this is the breakdown of the majors on your campus and that every person only has one major:

• 30% Business

• 15% Psychology

• 15% Engineering

• 10% History

• 10% Communications

• 5% Biology

• 5% Chemistry

• 5% Art History

• 5% Math

Following the steps listed above, you have defined your population (all the students on campus) and your strata (the aforementioned majors). Next up is separating the population into their strata. Hopefully, in this example, you would be able to access a database from the school to easily sort people into their majors.

Fg. 2 The business majors. pixabay.com.

After that, you and your professor determine the sample size of the entire study to be 2,000 students (it might sound like a lot but you go to a big college and you will have lots of help completing the study). Since you know the percentages of majors and how many subjects you want in total, you do the math to determine how many students from each stratum will be needed for the study. You’ll need 600 business majors, 300 of each psychology and engineering majors, 200 of history and communications majors, and 100 of biology, chemistry, art history, and math majors.

Then, you do random sampling from within the strata to reach the desired number of students from each major. All these students who have been stratified will then become your study participants.

Advantages of Stratified Sampling in Psychology

One of the most important advantages of using stratified sampling in psychology is that it ensures that there is an accurate representation of the population's demographic characteristics. If a researcher were just using random sampling of the population at large, they wouldn’t be certain that the subjects they had accurately represented the population that they were studying.

This also reduces sampling bias. Sampling bias happens when certain subsets of the population are more likely to be used as subjects for a study, resulting in an overrepresentation of that stratum. When the population is stratified before it is randomly sampled, it reduces this phenomenon.

The third advantage of stratified sampling is that it has good generalizability outside of the study. Since stratification means that people are actually and accurately represented (rather than studying race and randomly sampling people in Maine). This allows the results to be generalized to broader populations.

Disadvantages of Stratified Sampling in Psychology

A major disadvantage of this form of research is the time it takes to stratify. Researchers have to include extra steps such as defining strata and separating the entire population into the strata. These strata need to be extremely clearly defined so that there is no overlap between them. This can be time-consuming for researchers and add days to an already lengthy study.

Additionally, there are certain stratum that would be more difficult to organize people into than others. If you were separating people into shoe size, there would not be much difficulty since everyone wears the one shoe size. However, defining the strata becomes more complicated when there is crossover between the categories.

Additionally, even though it is better at reflecting demographics than just randomly sampling, it still does not accurately represent the entire diversity of the population.