# Is the cluster experimental design the same as the block experimental design?

Is the cluster randomised controlled trial the same as the randomized block design? They seem to both have the same goals.

Clusters are usually groups that have equal variability to the overall population. If I'm doing market research to bring a new cereal onto the market, it would be expensive to manufacture the product and ship it all over the country to see if everyone likes it. Instead I'll use a "test market", such as a few suburbs of a few cities across the country. I don't expect these places to have significantly different reactions to that of the overall population -- they're just geographically convenient.

If I divide a population into clusters and then randomly pick some of them, that's a cluster randomized trial. I won't use the whole population, only the clusters that get chosen.

Blocks are chosen to reduce variability. If I know that men and women will react to medications differently, then I want to ensure that my control group and my other groups have both men and women represented in them, and I'll want to compare the men against the men and the women against the women, so their reactions aren't going to be a confounding factor within my results.

If I divide a set of subjects into blocks, and then randomly assign treatments within the blocks, that's a randomized block design. I already have my sample by the time I'm dividing them into blocks, and every subject will get used.