Showing Cause & Effect With Experiments
Running an experiment requires that there be a treatment condition and a control condition. (A control condition is a situation in which subjects do not receive the treatment.) Having both conditions allows us to compare what happens with and without the treatment. If we cannot make this comparison, we cannot be sure that Treatment X really makes any difference for Behavior Y.
In addition, running an experiment requires that people are assigned randomly to the treatment and control conditions. If the assignments are not random, other variables besides the treatment may influence the results. For example, if we put all men in the treatment group and all women in the control group, any differences in their behavior could be caused by the difference in their sex and not by the treatment. When individual subjects are assigned randomly, their various personal characteristics get equally distributed in the treatment and control conditions, and so these characteristics do not influence the results of the experiment. For example, if we use random assignment, the average IQ of students in the treatment condition is equivalent to the average IQ in the control condition, so students in the treatment condition would not get higher test scores simply because they are more intelligent.
Well-designed experiments have another safeguard: the observers who collect the data are "blind." This means that the people collecting data do not know whether subjects are in the treatment condition or the control condition. If the observers are aware that subjects have received a treatment, the observers may have certain expectations about the subjects' behavior. This may lead to a "self-fulfilling prophesy." Because of their expectations, observers may behave differently toward those subjects and, without any desire to do so, may encourage certain behaviors or discourage others. Literally hundreds of experiments have shown that people who expect something to happen unintentionally make it happen. Researchers avoid this problem by not letting data collectors know which subjects receive a treatment.
When people test ideas about the causes of behavior, the three most common mistakes are:
(1) lack of a control condition,
(2) lack of random assignment,
(3) lack of "blind" observers.
Studies that have these flaws do not show that a treatment causes changes in behavior.
To return to the discussion of claims about HeartMath, click here.