psychology

Experimentation enables a person to focus on the possible effects of different factors. By alternating and controlling these factors, varied results can occur that will lead to data indicating the cause of behavior.

Demand characteristics in experiments

Most participants in experiments want to be "good subjects." Martin Orne demonstrated this some years ago when he attempted to identify a task people would do under hypnosis but not in the normal waking state. He had difficulty finding a task "waking" subjects would refuse to do. For example, he gave subjects 2000 sheets of paper containing simple addition problems. Each sheet required 224 calculations. Subjects were told to continue working until the experimenter returned. After 5.5 hours, they were still at it!

Orne made the task even more meaningless. After completing each sheet, subjects were to tear it up into 32 pieces, then go on to the next. Again, the subjects worked for several hours. In the context of the experiment, they attributed meaning to the task. Many believed it was a test of endurance. Orne also found that subjects would perform dangerous tasks such as picking up a poisonous snake barehanded, retrieving a coin from a fuming "acid" solution, and throwing a container of "acid" into the face of an experimenter's assistant.

Because of their desire to be "good" experimental subjects, people may be particularly sensitive and responsive to any cues that indicate what behavior is expected. These cues, or demand characteristics, may represent uncontrolled variables in an experiment. Examples include campus rumors about the experiment, the arrangement of the equipment used, and characteristics of the experimenters, such as their age, sex and race. It has been shown, for instance, that white subjects may express less prejudice in the presence of a black experimenter. Similarly, a male experimenter may smile more at female than at male subjects, and thus inadvertently contribute to a sex difference in behavior. To minimize the effect of demand characteristics, experimenters typically standardize their instructions and read or even tape record them.

Similarly, the expectations of the experimenter may contaminate the results; this is known as experimenter bias. Investigators have hypotheses about how subjects will respond, and may unintentionally treat subjects and the experiment in a way that confirms their expectations. Robert Rosenthal demonstrated this self-fulfilling prophecy some years ago. In his study, one group of student experimenters thought subjects would interpret photographs of people's faces as reflecting success, while others thought their subjects would see failure. Results were consistent with the experimenters' expectations. Clearly, they had unintentionally communicated their expectations to their subjects.

The use of a double blind design can help to reduce the effect of these demand characteristics and the effect of experimenter bias. By arranging the experiment so that keeping one individual who knows which condition every subject is in and keeping both the subject and the person directly administering the test unaware of the independent variable, the results can be protected from bias.

For example, in medical research all of the subjects in a new drug trial are given a pill. Some of the subjects receive the real pill, which contains the new medicine. Other subjects receive a placebo, a pill that looks like the new drug but which contains only sugar. The new drug and the placebo look so similar that even the doctor administering the drug can't tell the difference. The only person who knows which subjects are receiving the real medicine and which are receiving the placebo is the hospital administrator who is overseeing the experiment. That way, the results can be analyzed objectively. If greater improvement is seen in the group that received the real medicine, then it can be safely concluded that the medicine worked. This method also helps us determine the side effects of the drug.