size of control group v.s. size of treatment group

bowenz

Why the size of control group and that of treatment group are slightly different? Why not use the same group size? I find it in chapter 1.1 the case study and also the exercises for it.

David

Hi, the answer is "it depends". I'll talk about it in terms of a clinical trial and patients to ease the description, and this is a very incomplete list:

1. Patients might be randomized to groups as they arrive, and the total sample size wasn't decided on ahead of time. This is particularly likely for a study where recruitment runs over months or years as part of a long-running study or as part of a study focused on a rare condition, but it can also be possible with short-term studies where it is convenient to allocate people to their group right away.

2. Sometimes patients stop participating in the study, so the original sample size at the start may differ from the end of the study. In cases like this, other checks are likely to be (should be) done in case there are reasons for patient withdrawals that relate to the experiment itself.

3. Sometimes having a patient in treatment is more expensive than having a patient in the control group (or vice-versa). In this case, it might be advantageous to design an imbalance into the study, as most studies are at least partially constrained by financial considerations.

4. If there are multiple treatment groups and a single control group that is used as a baseline for all of the treatment groups, it can be useful to allocate more users to the control group than are allocated to any single treatment group. This strategy can reduce the estimated effects for all groups.

This is an incomplete list, and perhaps you can come up with additional practical considerations now that you've got these as examples. :)

David

Register to Add a Comment