BLIND EXPERIMENTS


"BLIND" EXPERIMENTS

In experiments, as in surveys, you must not bias your observations or treatments with your own opinions or preconceptions about which group or treatment should yield better results. Some events, of course, are not disputable, such as the fact that a rat has died. However, when making observations that are not as clear-cut , such as assessing the happiness of a person's marriage , it is all too easy to let unreliable judgment creep in ” even though you are trying to be objective and "scientific."

Not only you as an experimenter but also your subjects ( especially if they are humans ) can influence the outcome of an experiment without even trying. An example of a biasing influence is the placebo effect , a well-known effect in medical research. A placebo (such as a brightly colored pill that has no real effect) and a pep talk from a sympathetic physician are enough to cure many ailments. In an experiment on alertness, for example, if students believe that the vitamin supplements they get with their math lessons are intended to make them less sleepy during class, they may actually feel more alert (or more drowsy if they have a bias against the experiment's success). In an experiment on anxiety, if the patients believe that the pill they are getting contains a drug with a powerful relaxing effect, they will feel more tranquil than if they believe that they are just getting breath mints.

The placebo effect can occur in many kinds of experiments, not just in medical research. To avoid the effect, you should prevent subjects from knowing which experimental group they are in, and you should not tell them anything about the expected results. Keep them "blind" as much as possible. Ethical considerations require that they know about any risks and that they give "informed consent ." However, you can still design the treatments to avoid biasing the results. For example, if one treatment requires a group of people to take pills, make sure that all of the other groups get pills too, even if they are just sugar.

The people who record the experimental results should also be unaware of the assignment of subjects to groups. They too should be "blind." Make sure they know exactly what to measure, such as weight without clothes, learning time to the nearest second, or anxiety on a particular scale. But avoid explaining more than they need to know. If you satisfy their curiosity by explaining what is going on while the study is in progress, you will never be sure whether they unconsciously affected the results. Explain the issues after the study is complete. You do not want anyone 's prejudices to influence the measurements. Even if you are making the observations yourself, you can still keep yourself blind by not knowing which subject is in which experimental group. Have an assistant assign the subjects randomly to the various groups, leaving you pure and untainted.

Medical studies are often characterized as single blind or double blind. When only the subjects do not know which groups or treatments they have been assigned to, the experiment is called single blind . When both the experimenter and the subjects are kept unaware of the assignment, the study is called double blind . Double blind studies are the most reliable.




Six Sigma and Beyond. Statistics and Probability
Six Sigma and Beyond: Statistics and Probability, Volume III
ISBN: 1574443127
EAN: 2147483647
Year: 2003
Pages: 252

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net