Does 'control' in controlled experiment refer to:

  1. Having a control group for comparison, i.e "an experiment or trial that uses controls, usually separating the subjects into one or more control groups and experimental groups." Source

  2. Controlling for extraneous variables, i.e "A controlled experiment is simply an experiment in which all factors are held constant except for one: the independent variable." Source

  3. Both 1. and 2.

Searching for the definition of a controlled experiment was not very helpful in clarifying this as seen from the different sources. They are related concepts but still, they are not the same thing.

Related to this question is what then qualifies as a controlled experiment? If I conduct an experiment with a control group and treatment group with differences in the independent variable, but do not control for extraneous variables (badly designed), would this still be considered a controlled experiment since there are factors which are not held constant? Do all controlled experiments require a control group?


1 Answer 1


Control means that the experimenter randomly assigns the experimental conditions to all subjects and makes sure that there are no systematic differences other than the experimental conditions. This is closely related to (1), however the term control group has usually a specific meaning in clinical research while experimental condition has a broader meaning and can also refer to non-clinical studies.

The idea is that by means of this procedure you can also control for extraneous variables (related to (2)) given sufficiently high sample size because due to the random assignment of experimental conditions there should not be any bias and all differences with respect to other variables should cancel each other out across subjects. So any differences you observe between experimental conditions can be attributed to these experimental conditions.

This is often contrasted to correlational research where variables are just observed but not assigned (i.e., not controlled) and thus there might be confounding third variables.

So, an experiment qualifies as controlled if you randomly assign all experimental conditions and make sure there is no systematic difference in how subjects are treated apart from your experimental conditions. For example, you decide randomly who out of group of patients gets a treatment and who is in a control group (or a different experimental group). If there was any non-randomness in this assignment (e.g., patients decide for themselves or treatment is decided based on severity of illness), this would not be a controlled study. If all treatment patients are tested in the morning and all control patients are tested in the afternoon, there would be a systematic difference between both groups other than your actual treatment and your study would not be controlled. However, patients might differ with respect to many other variables (they do not need to be identical twins), but these differences are not systematic.

  • $\begingroup$ I'd argue randomness isn't a requirement for control - what you're really seeking is balance. When assigning patients to treatment arms in a clinical trial, random assignment could yield differences in age or disease severity by chance alone. Although this is unlikely, it's better to explicitly balance relevant patient characteristics between arms to preclude the possibility entirely. As long as the patient characteristics are balanced between arms, it's a proper control regardless of if you assigned them randomly or specifically to achieve balance. $\endgroup$ Feb 18, 2021 at 19:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.