What Is Stratified Random Sampling?

What Is Stratified Random Sampling?

Let me explain stratified random sampling to you directly: it's the process where I divide a population into smaller subgroups, or strata, based on shared attributes like age, gender, income, or education. These strata ensure that members with similar characteristics are grouped together, and then I take random samples from each one.

You might hear it called proportional random sampling or quota random sampling, and it's useful for things like studying demographics or life expectancy in a population.

How Stratified Random Sampling Works

When you're dealing with a large population that's hard to study entirely, you turn to sampling to make it manageable. I select a smaller group, the sample, to represent the whole population.

In stratified random sampling, I divide the population into homogeneous strata and then pick random samples from each. This lets you analyze differences across demographic groups.

It's important to note that stratified sampling highlights group differences, unlike simple random sampling where everyone has an equal chance.

For example, if I'm researching MBA graduates getting job offers, I'd strata by gender, age, race, nationality, and career background, then sample proportionally from each to pool into a final sample.

Simple vs. Stratified Random Samples

Both simple and stratified random samples are tools for measurement, but stratified divides the population into strata based on characteristics, making it more complex and potentially costlier than simple random sampling.

Use simple random sampling when there's limited population info, too many differences for clean subsets, or only one key characteristic.

Take a candy company studying 10,000 customers: a simple random sample of 100 works if strata differences aren't significant, but stratified would divide by age or income if needed.

Proportionate vs. Disproportionate Stratification

Stratified sampling ensures subgroups are represented, and it can be proportionate—where each stratum's sample size matches its population proportion—or disproportionate, where it's not.

Proportionate is more precise, like in a study of 180,000 people sampled to 50,000, calculating stratum sizes with the formula (sample size / population size) × stratum size.

In disproportionate, you might over-sample certain groups based on needs, but remember, no overlapping strata—each person fits in one only to avoid bias.

Advantages of Stratified Random Sampling

The key advantage is capturing population characteristics proportionally, like a weighted average, which reduces estimation error and boosts precision over simple random sampling, especially with diverse strata.

Disadvantages of Stratified Random Sampling

You can't use it if you can't identify and classify every population member into one subgroup without overlaps, which could lead to misrepresentation if sorting is too hard.

Example of Stratified Random Sampling

Suppose I want the GPA of U.S. college students from 21 million; I'd sample 4,000, strata by majors like English or science, then adjust to proportional representation for accuracy.

When Would I Use Stratified Random Sampling?

Use it when you need insights on subgroups like race or gender in the population.

Which Sampling Method Is Best?

It depends on your analysis; simple is easier, but stratified gives more accurate results for diverse data.

What Are the 2 Types of Stratified Random Sampling?

Proportionate samples strata proportionally; disproportionate over- or under-samples based on study design.

How Are Strata Chosen for Stratified Random Sampling?

Strata come from shared characteristics in your population, like gender or age.

The Bottom Line

Stratified random sampling creates subgroups by factors like age or income, sampling from each to represent under-represented groups better than simple methods, though it requires good population data.