Stratified random sampling: What it is and how to use it

Let’s say you need to gather information about a large population, like the residents of a major city. It might be impractical — not to mention expensive — to survey every person in that population. That’s why researchers often select a smaller subset of the population, or sample, to represent the whole.

There are a number of ways to select a sample to make sure it’s most likely to be representative of the larger group. One of these techniques is stratified random sampling.

In this article, we’ll explain what stratified random sampling is, share a step-by-step process for how to conduct stratified random sampling, review its advantages and challenges, discuss when to use the method, and share some examples of stratified random sampling in practice.

A brief explanation of stratified random sampling

Stratified random sampling is a data-collection method that involves dividing a population into smaller subgroups, called strata, that share similar characteristics or attributes, such as income, education level, age, race, or gender.

The final sample is made up of randomly selected members of each stratum. To analyze the survey results, researchers compare the responses of the subgroups from each stratum to each other. This sampling method is also known as proportional random sampling or quota random sampling.

Steps to conduct stratified random sampling

Here’s a brief overview of the steps for conducting this type of sampling:

  1. Define your total population of interest. For example, your population of interest might be the residents of Los Angeles, California, or college students in the United States.
  2. Determine the strata for your sample. For example, you might select race, gender, income level, education level, nationality, or age group. Each member of the population can only be in one strata.
  3. Define the sample size for each stratum. To determine the sample size for each stratum, first divide the population of the stratum by the total population. Then multiply that number by the total sample size you need for your survey. The ratio of the sample size for each stratum to the population of the stratum should be the same as the ratio of the stratum population to the total population. That way, the responses of each group will be weighted according to the group’s size relative to the population.
  4. Select a random sample from each stratum or subgroup. Once you’ve determined your subgroups, randomly select participants from each stratum. You can do this by using probability sampling methods such as simple random sampling or systematic random sampling. With probability sampling, every member of the population has an equal chance of being selected.
  5. Review stratum results. You’ll want to verify a final sample and make sure each participant of the population belongs to only one stratum and participants don’t overlap with another stratum or subgroup.
  6. Consolidate all stratum samples into one representative sample. This will ensure you have an accurate representation of the total population of interest.
  7. Conduct the survey with the chosen subgroups.

Stratified random sampling examples

Let’s look at a few simple examples of stratified random sampling:

  • Researchers might seek to compare the marital status of women with a certain education level with that of men with a similar education level. This could be used to understand the effect of education levels on the likelihood of marriage for both genders. In this case, gender is the selected stratum, with random samples from each subgroup selected in equal proportion.
  • Researchers might want to study the average GPA of college seniors in Oregon with different majors. They could randomly select 1,000 students out of 100,000 and use the students’ majors as the subsets.
  • A researcher could want to study wage differences between genders. The researcher could stratify a larger population according to pay grades and randomly select a sample across each of the pay grades. The researcher could then compare the samples using gender to investigate wage gaps.

Advantages of stratified random sampling

Stratified random sampling offers several key advantages:

  • Less chance of bias: Since researchers randomly select a sample from each stratum, there is less chance of bias in the sample selection.
  • Greater efficiency: Because a population is organized into groups with similar characteristics, data collection and analysis are more efficient, resulting in time savings.
  • Lower cost: Studying smaller groups with similar characteristics means researchers don’t have to spend money to survey every member of a larger population.
  • Improved data quality and accuracy: When members of the subgroups are homogenous relative to the larger population, stratified random sampling can provide greater accuracy and precision. Stratified random sampling produces characteristics in the sample that are proportional to the total population of interest.

Situations that call for stratified random sampling

Stratified random sampling is a good methodology to use to gain insights into strata or subgroups within a larger population — for example, when the research seeks to explore differences among groups based on gender, income level, race, education level, age, and so on.

It can also help to collect more accurate data, because different strata are represented in the sample according to their representation in the population as a whole.

Let’s look at a few potential use cases of stratified random sampling:

  • A medical research study could examine the prevalence of a disease in different age cohorts, with the total population of interest stratified by age brackets — such as 18–29, 30–45, 50–65, and 66-plus.
  • A health insurance group seeking to design a healthy lifestyle campaign to fit the needs of its target audience could stratify subgroups by factors such as fast-food consumption, income level, whether participants exercise regularly, and so on. The data the survey collected would then inform the most accurate healthy lifestyle messaging for each subgroup or strata to help them achieve their health goals.
  • Let’s say you own a restaurant with a reputation for being the ideal place for affluent customers in their 60s, but you want to attract more of the 30–40-something crowd who prefer more casual dining. Your population of interest could be anyone in the younger age group who has dined at your restaurant in the past six months. Your subgroups could be made up of
    • Those who are 30–35, 36–40, and 41–45
    • The frequency at which these groups dine outside the home at restaurants like yours
    • Menu options they might find appealing, ranked by preference

The data you collect would reveal menu preferences from each age group and the frequency of dining out, which you could use to create new menu options, along with messaging to appeal to the 30–40-something crowd.

  • An academic researcher would like to determine the number of students with a bachelor’s degree in history who received a job offer upon graduation in 2020. In our example, let’s say there are 100,000 graduates with a bachelor’s degree in history in 2020.

The researcher could divide that population into strata, such as age, gender, and race, and select a random sample from each stratum according to the percentage of the total population of interest that stratum represents. They would then combine the subsets of the strata to form a random sample.

  • A researcher would like to examine opinions about religion for different age groups in the United States. Instead of gathering data from all U.S. citizens, they could collect random samples from 10,000 citizens stratified by age, like this: 18–29, 30–39, 40–49, 50–59, and 60 and above. Each subgroup would have various characteristics, including education level, income level, nationality, race, and so on.

Challenges and other considerations of stratified random sampling

As beneficial as stratified random sampling can be, there are some drawbacks:

  • In order for stratified random sampling to work, you must be able to identify every member of the population you’re examining, and you must be able to classify them into only one subgroup, which can be challenging.
  • This method can be time-consuming. You’ll need to find and review a population of interest and develop thorough information about each category or subgroup that participants should be divided into.
  • You have to be careful to avoid overlap in your sample, where participants may fall into more than one subgroup or strata. If this happens, your results will be less accurate.
  • It’s not ideal for all types of studies, such as when you don’t have consolidated information on subgroup characteristics.

A comparison of stratified random sampling with other sampling methods

Here’s an overview of how stratified random sampling differs from other methods.

Stratified random sampling vs simple random sampling

Simple random sampling selects a smaller group or sample from a larger group of total participants or population. This approach ensures that each participant has an equal chance of being chosen.

Researchers use simple random sampling when they want the data they collect to be representative of the total population of interest. It’s also a good method for when researchers must choose samples quickly, such as with opinion polls and market research, and when the overall population size is small.

Stratified random sampling vs systematic sampling

Systematic sampling is a probability sampling method in which members are chosen from a larger population based on a random starting point and selected at a fixed, regular interval. One example would be choosing every 10th person on a list of all members of the population. The interval is determined by dividing the population size by the desired sample size.

Researchers may use systematic sampling when they have a restricted budget and need a simpler process. With systematic sampling, there’s no need to create multiple samples, which can be costly and labor intensive. This method is also useful when researchers need results more quickly, since it only requires a small part of the total population being sampled.

Stratified random sampling vs cluster sampling

With cluster sampling, researchers divide a larger population into groups known as clusters, such as by cities, schools, or geographic location. Researchers then randomly select among the clusters to create a sample. With stratified random sampling, you choose some individuals from all groups, but with cluster sampling, you randomly select entire groups and include all units of each group in the relevant sample. In other words, the sample is the entire cluster.

By its nature, cluster sampling is more affordable to conduct, while stratified random sampling is more accurate and precise. Researchers may use cluster sampling when budget is a concern, when it’s difficult to collect data from an entire population, or when there’s a need to study large, geographically dispersed populations where interviewing each subject isn’t an option.

Surveys made easy with Jotform

Stratified random sampling is just one of several data collection sampling methods. If you have a large population to survey, it isn’t realistic to connect with every member of that population. That’s when stratified random sampling is especially helpful. And, once you identify the strata or subgroups you want feedback from, you can use Jotform questionnaires and surveys to easily collect and consolidate that data.

Photo by Photo By: Kaboompics.com

AUTHOR
Kimberly Houston is a conversion-focused marketing copywriter. She loves helping established creative service providers attract and convert their ideal clients with personality-driven web and email copy, so they can stand out online, and get more business, bookings, and sales.

Send Comment:

Jotform Avatar
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Podo Comment Be the first to comment.