This is the basic strategy behind an experimental epidemiological study. To show that a treatment is effective, the improvement rate in a treated group (T) should be greater than the improvement rate in the untreated group (NT), and that the differences in rates should be unlikely they are due to chance.

The control group (NT) should be similar to the treatment group (T) in all characteristics associated with the expected outcome (recovery), and should not receive the studied treatment. The control group may be receiving the prevailing, or current standard of care, or a placebo.