Last week we saw how some experiments incorporate control groups. In a pharmaceutical experiment, for example, an experimental group may take a pill containing the drug being evaluated, while a control group takes a pill which contains an inert substance. The pill with the inert substance is called a placebo, and an improvement in the control group similar to the improvement expected in the experimental group is called a placebo effect.
What causes placebo effects? In medical research it is often difficult to tell. In behavioural research, though, they are often due to motivation; motivation can also affect medical studies. For example, people taking a pill they think will help them lose weight may get enthusiastic and decide to go on diets as well. Even if they're actually taking placebos, the diets will still get them to lose weight.
The value of the control group is readily apparent here. Weight loss due to the drug is not confused with weight loss due to other factors. If there were no control group and the drug were ineffective it would be dangerously easy to conclude that the drug was effective.
There is also the phenomenon of what is known as the experimenter effect. People taking part in an experiment often want to help the experimenter, and they do it by behaving as they think the experimenter would like to see them behave. Again the control group eliminates the danger of mistaking helpfulness for the effect the experimenter is looking for.
One form of the placebo effect is called the Hawthorne effect, after the Hawhorne, Illinois, plant of the Western Electric company, where it was first noted. The workers at the Hawthorne plant underwent a series of improvements in their working conditions to see if productivity could be increased. The researchers then successively removed the improvements which seemed to increase productivity to confirm that they did in fact affect production. They of course expected productivity to decline, but found instead that it continued to increase.
Again the effect was motivational. The conventional explanation is that the workers interpreted all the changes as signs of concern about their working conditions, and worked harder to show their appreciation. Of course, maybe they were just trying to keep the company from making things worse than they had been to begin with. Either way, the effect was motivational.
The existence of effects like these makes controlled studies indispensable. Researchers' habitual dismissal of anecdotal studies comes from their awareness of this problem. Anecdotal studies are important – for one thing, if you don't get an effect in anecdotal reports you know you can dispense with a controlled study – but when they appear to demonstrate effects they must be confirmed with controlled studies.