Muddy Points

Lesson 1: Data collection

Fall 2025

1. Cluster sampling versus stratified sampling?

There are two main differences between cluster sampling and stratified sampling:

1. How groups (or clusters) are formed
2. How observations/individuals are selected within each group

I think most of us were confused about the first point. Whats the difference between a subgroup in stratified sampling and a cluster in cluster sampling?

In stratified sampling, subgroups (or strata) are formed based on shared characteristics relevant to the study. Typically, the individuals within the groups/strate are more homogeneous than individuals from other strata. The goal is to ensure that each subgroup is represented in the sample.

In cluster sampling, clusters are often formed based on natural groupings or locations. Typically, the individuals within each cluster are diverse. The goal is to randomly select entire clusters.

We’ll discuss more in class when we look at the example!

2. When to use Sakai vs when to use the class website?

The class site will contain all course information. The only time you need to use Sakai is for submitting homework assignments, submitting quizzes, and checking your grades.

Fall 2024

Our long delayed example of cluster sampling, stratified sampling, and multistage

This answer comes from our sweet, sweet friend, ChatGPT:

Here’s a set of examples focused on studying patient satisfaction within a single hospital:

1. Cluster Sampling:

Example: A researcher wants to assess patient satisfaction within a large hospital but cannot survey every patient. The researcher could:

• Divide the hospital into floors (clusters).
• Randomly select a few floors.
• Survey all patients on the selected floors.

2. Stratified Sampling:

Example: The researcher wants to ensure representation of patients from different types of care (e.g., inpatient, outpatient, emergency). The researcher could:

• Divide the patients into strata based on the type of care they received.
• Then, randomly select patients from each care category to ensure proportional representation in the sample.

3. Multistage Sampling:

Example: To study patient satisfaction within the hospital, the researcher could use multistage sampling:

• First stage: Randomly select departments (e.g., Oncology, Surgery, Cardiology).
• Second stage: Randomly select wards or units within each department (e.g., ICU, general ward).
• Final stage: Randomly select patients from each unit to survey about their satisfaction with the care received.

This example is specific to a single hospital, using different sampling techniques to focus on patient satisfaction within various departments and care types.

Now back to Nicky

One of the key pieces of info here is that we are sampling patients! In the cluster sampling, we survey all patients in our randomly chosen cluster. In stratified, we set the subgroups, but we only sample some of the patients in each subgroup!

In multistage, there is a step where we randomly chose different clusters (like the department or wards) THEN we randomly select patients within those chosen clusters. Each stage can be identified with a different sampling technique (outside of multistage). Can you map each stage to their respective sampling method?