secular-knight 69

"A double-blind test is a control group test where neither the evaluator nor the subject knows which items are controls. The purpose of double-blind testing is to reduce error, self-deception and bias"

If neither the subject, nor the evaluator knows which are the controls, how does one find out what the controls were?

Wounded King

Because there are careful records kept which allow you to discern which were controls after the experiment.

AtomSmasher

Often, the persons doing the evaluating intentionally or unintentionally taint the results because they know who received the treatment and who did not. To avoid this, the evaluator does not know who was or was not treated. After the results are gathered, the identity of the control and test subjects is uncovered so the data can be compared without any possible bias by the test subjects or researchers.

Undercurrent

One way to do it is to have an independent experimenter sort pills (say) and placebo into containers. The containers have a unique number on them and the independent part records which numbers have real pills and which have placebos.

The actual person who's giving the pills out to the subject, records the number on the container that each subject gets, without having ever seen the record of which numbers are real and which are placebo. After the experiment is done, the two rcords can be put together to determine which subjects got real pills or not.

secular-knight 69

Thank you for your responses.

Coragyps

A friend of mine was one of the experimenters in a double-blind study to look at a replacement candidate for Thorazine back in the 70's. The subjects were inmates at an asylum, and were on Thorazine before the study to keep them "subdued." My friend would help pass out the coded pills: placebo, Thorazine, and new stuff. He said that 45 minutes later the folks that got the real article were razzing the others - "You got the sugar pills!" and "You got the new crap!" - and turned out to be correct.

AtomSmasher

That's hillarious, good thing it wasn't a cancer drug

doghouse

An additional point. You don’t always wait until the end of the study to start analyzing the data, at least in medical studies. For example, when testing a new treatment, people other than those doing evaluating the subjects will often look at the results at a certain time before the planned study is completed. If there are statistically significant differences at that point, they may stop the study. For example, if analysis shows the treatment is obviously detrimental after only half the patients have enrolled, they may stop the study because of ethical concerns. The same if the treatment shows an obvious benefit.