You often hear studies described as randomized, double blind, and crossover. That's supposed to be a good thing, but what exactly does it mean?

First of all, in a randomized study, the subjects are assigned at random to the experimental group and to the control group. If the assignments weren't random, the experimenters might unconsciously assign, for example, agreeable people to the test group and disagreeable people to the control group. That would influence how each group responded in the study.

In an open trial, both the researchers and the subjects know the full details of the treatment. These trials are vulnerable to the placebo effect; that is, because the subjects believe that the stuff they're testing should work, it does work. Sometimes an open trial is necessary; for example, it's pretty hard to conceal the details from a subject when they're getting a surgical treatment.

In a single blind experiment, the subjects do not know whether they are the test subjects or belong to the control group, but the experimenters do know. A single blind design is used when the experimenters must know the full facts in order to carry out the experiment. Because the experimenters know what's going on and have expectations about possible results, there is a risk that the subjects might respond to experimenter's bias: that is, the experimenters might give subconscious hints that influence how the subjects react to the experiment.

In a double-blind experiment, neither the subjects nor the researchers know who belongs to the control group and who belongs to the experimental group until long after the data is gathered. This lessens the influence of the scientists' expectations and unintentional physical cues on the subjects, helping to eliminate the placebo effect and experimenter's bias.

In a crossover study, each patient is given the test medication for a time and then the placebo for a time, in random order. A crossover study minimizes the variability between patients because each patient crossing over in effect serves as his or her own control.

As you can see, a randomized, double blind, crossover study is the best bet for achieving results that are untainted by the preconceptions and biases of both experimenters and subjects. That's why it's known as the gold standard of experimental design.