Cluster vs. Stratified Sampling- AP Statistics Exam Review
What the Heck Is Sampling Anyway?
Before we dive into the differences, let's get one thing straight: sampling means picking a subset of a population to study. You can't survey 300 million people, so you pick a smaller group and use it to make inferences about the whole.
The AP Statistics exam loves testing whether you understand how to pick that subset. Cluster and stratified sampling sound similar but work completely differently. Mixing them up will cost you points.
Stratified Sampling: Divide and Conquer
In stratified sampling, you split your population into separate groups called strata, then randomly sample from each stratum.
The key word here is each. Every single group gets represented. You're not skipping anyone.
How It Works
- Identify the strata (groups that share a characteristic)
- Randomly select individuals from within each stratum
- Combine all the samples
Real Example
You want to survey students about lunch preferences. You divide the school into freshmen, sophomores, juniors, and seniors. Then you randomly pick 50 students from each grade. Every grade is represented proportionally.
Cluster Sampling: Pick Groups, Not Individuals
Cluster sampling is different. You divide the population into clusters, then randomly select entire clusters and survey everyone within them.
The key difference: you're not sampling individuals from each group. You're picking whole groups and including all members of those groups.
How It Works
- Divide population into clusters (usually geographic or naturally occurring)
- Randomly select some clusters
- Survey every member of the selected clusters
Real Example
Using the same school scenario: you divide the school into homeroom classes. You randomly select 5 homerooms. You survey every student in those 5 classes. Students in the other homerooms? Not included.
Side-by-Side Comparison
| Feature | Stratified Sampling | Cluster Sampling |
|---|---|---|
| Division method | By similar characteristic (strata) | By geographic or natural grouping |
| Selection unit | Individual members | Whole clusters |
| Who gets surveyed | Some from every stratum | Everyone in chosen clusters |
| Variance | Lower (more precise) | Higher (less precise) |
| Cost | Higher (travel to many locations) | Lower (concentrated locations) |
How to Identify Which Method on the Exam
The test writers will describe a scenario. Your job is to figure out which sampling method is being used. Here's the mental shortcut:
If the description mentions selecting individuals from within groups → stratified.
If the description mentions selecting entire groups and including everyone in those groups → cluster.
Watch Out For Confusing Language
Read carefully. "Randomly selecting 3 homerooms and surveying every student in those rooms" is cluster sampling. "Randomly selecting 20 students from each homeroom" is stratified sampling.
The difference is whether you're sampling individuals or including whole groups.
Common Exam Mistakes
- Thinking strata and clusters are the same thing. They're not. Strata are homogeneous groups (members are similar). Clusters are heterogeneous (members are different, like a mini-version of the whole population).
- Confusing stratified with stratified by convenience. Just dividing into groups doesn't make it stratified. You must randomly sample within each group.
- Forgetting that cluster sampling often has higher variability. If clusters are too similar to each other, your estimates become less accurate. The exam might ask about this.
Quick Reference for Test Day
When you see a sampling question, ask yourself:
- Is the population divided into groups that are similar within and different between? → Stratified
- Is the population divided into groups that are mini versions of the whole population? → Cluster
- Are you selecting individuals from each group or whole groups? → This tells you which method applies
That's it. The distinction is straightforward once you stop overthinking it.