A researcher wants to know whether a new drug reduces blood pressure. She gives the drug to one group and a placebo to another. She measures blood pressure in both groups. The only difference between them is the drug. That is an experiment: a research method where variables are manipulated to determine cause and effect.

The key feature is control. By randomly assigning participants to conditions, the researcher can be confident that any difference in outcomes is caused by the manipulation, not by pre-existing differences. Random assignment is what separates a true experiment from a correlational study. Correlation shows that two things vary together. Experiment shows that one causes the other.

Experiments have limitations. They can be impractical, unethical, or artificial. You cannot randomly assign people to smoke cigarettes. You cannot easily study long-term effects in a laboratory. And even well-designed experiments can be biased by demand characteristics, where participants guess the hypothesis and change their behavior. The double-blind procedure, where neither the participant nor the experimenter knows who got the treatment, is one way to reduce that bias. The method is powerful, but it is not magic.