# Stratified Random Sample vs Cluster Sample

Updated: Oct 13, 2019

For starters, students need to understand the most fundamental idea of good sampling: the simple random sample (SRS). Hopefully you used the Beyonce activity to introduce this concept, but let’s realize that the SRS has some limitations. When taking an SRS of high school students in your school, isn’t it possible that your whole sample might all be Freshman? All Seniors? Also, it might be very difficult to track down an SRS of 100 students in your high school.

So what is the solution? It could be a stratified random sample or a cluster sample.

## Stratified Random Sample

If there is a variable that you know will be closely associated with the response variable that you are trying to measure, a stratified random sample might help you make better estimates. For example, suppose you are trying to take a sample of 100 students to ask them whether or not they support a new parking lot at the school. Certainly Freshman students will have different support levels than Seniors. In this case, you might want to ensure that you are getting all four grade levels represented in your sample. Here are the steps for the stratified random sample:

Take the population of all high school students at your school and split them into groups (strata) by grade level. Within each strata, students are similar (same grade level).

Take an SRS of 25 students from each grade level.

We suggest using the Justin Timberlake activity to help students understand the advantages of using a stratified random sample.

**Cluster Sample**

Locating 100 different students within the school is quite time consuming. Instead of an SRS or a stratified random sample, you might want to use a cluster sample to make data collection easier. When setting up a cluster sample, it is important that each cluster is a good representation of the population. In the school setting, this means that each cluster has to have a good representation of all four grade levels. This happens in __some schools__ in a home room. We are talking about a school that intentionally has all four grade levels represented in each home room. Here are the steps for the cluster sample:

Take the population of all high school students at your school and split them into groups (clusters) by home room. Within each cluster, students are different (multiple grade levels).

Take an SRS of 3 of the home rooms. Interview

__all__students in each of the selected home rooms.

**What is the same for the two sampling methods?**

Both sampling methods take the population and split it into groups.

Both sampling methods utilize the concept of an SRS.

**What is different for the two sampling methods?**

The groups for stratified random sample are homogeneous. The groups for cluster samples are heterogeneous.

For stratified, one takes a sample from each group (strata). For cluster, one takes

__all__individuals from the selected groups. “Some from all” versus “all from some”.