When doing experimental research you take a random sample from the population in which you are interested. You then randomly select a control group and a treatment group. The treatment group is given the treatment that you are studying (this can be a trial for a medication, or a lifestyle change like exercising, etc...) The control group does not receive the treatment. (In the case of medicinal experiments, ideally the experiment is conducted double-blind; no one but the researcher, who does not interact with the test subjects, knows who receives the treatment.)
Since both the control group and the experimental group are randomly drawn from the population of interest they will share equally in the traits of the population. E.g. if 10% of the population smokes, then about 10% of the study group will smoke, and assuming random distribution into control and treatment groups 10% of each of these groups will also smoke.
Confounding variables are variables that affect the results of the study but which are not directly accounted for. Some well-known confounding variables that affect many studies are the placebo effect. the nocebo effect, and the Hawthorne effect.
Suppose you were testing a weight-loss supplement. Some confounding variables not mentioned might be exercise habits, whether or not you smoke, diet, genetic traits, etc... By randomly choosing your control and experimental groups, the groups should exhibit roughly the same backgrounds, propensities, etc... allowing you to look at the supplement's effect.
As you know, a control group in any experiment is essential. Without it, you cannot tell whether an experiment has been successful. Researchers have drawn attention, rightfully, to something called the placebo effect.
A placebo group accounts for the effects from treatment or experiement that do not depend on the treatment or experiment itself. From a methodological point of view, this is an important point, because without a placebo group to compare against, it is not possible to know whether the treatment or experiment had any efficacy. In this sense, the control group does account for variable. In fact, this is why a control gorup is essential to have.
With that said, here are a few example.
If a researcher wanted to see if watching TV hinders children from learning numbers in preschool, then he or she may give a test after a TV program. But without a control group, the researcher would never know if his or her research was valid without a control group. If the control did better or about the same, then the thesis of the experiment would not be valid.