Statistics How To

Cluster Randomization in Controlled Trials

Share on

Design of Experiments > Cluster Randomization

What is Cluster Randomization?

cluster randomization

A cluster is a group that has some characteristic in common.

In cluster randomization, groups of participants are allocated to treatment groups and control groups. The groups (clusters) have something in common, like:

  • Geographic location,
  • Age,
  • Gender.

The groups don’t have to be clusters of people. For example, they could be:

  • Community organizations,
  • Clinics,
  • Countries.

The groups are always pre-existing, rather than being created by the researcher.

Experiments that use cluster randomization are sometimes called cluster randomized trials (CRTs) or group-randomized trials.

When to Use CRTs

CRTs are often used to measure the effectiveness of management strategies and public health interventions. They are sometimes used in a clinical setting to study drug effectiveness (Mazor et. al, 2007). In general, they are a good choice when looking at decisions to treat or not treat entire groups (e.g. whether to immunize groups or provide mosquito netting for an entire geographic area) or when contamination risk is high. Contamination means that individuals are in close contact and influence (“contaminate”) each other. For example, if you want to study how diet affects prison inmates, this carries a high risk of contamination as inmates may swap or share food.

CRTs vs. Individually Randomized Trials

Trials that use cluster randomization are reported and analyzed in the same way as trials that randomize individuals. However, instead of analyzing the trial at the individual level, it’s analyzed at the cluster level. “Cluster level” just means that the results for the entire cluster is looked at, instead of one level further down—the individual level.

Cluster randomized designs usually have lower statistical power than similarly sized experiments that randomize individuals instead of clusters. They also tend to have a higher number of false positives (Type I errors). However, the design has many benefits including better cost efficiency, a lower risk of experimental contamination, and a lower risk of noncompliance from study participants.


Donner, A. & Klar, N. (2010). Design and Analysis of Cluster Randomization. Wiley.
Mazor, K. et. al. (2007). Cluster Randomized Trials: Opportunities and Barriers Identified by Leaders of Eight Health Plans. Med Care, 45:S29-S37.

Stephanie Glen. "Cluster Randomization in Controlled Trials" From Elementary Statistics for the rest of us!

Need help with a homework or test question? With Chegg Study, you can get step-by-step solutions to your questions from an expert in the field. Your first 30 minutes with a Chegg tutor is free!

Comments? Need to post a correction? Please post a comment on our Facebook page.