What is a matching group in a research?

Group matching attempts to use data that has been collected previously. In an idea situation, an experimenter could randomly assign people to groups. For example one randomly chosen group gets the drug, while another randomly chosen group does not. Real life is rarely so nice, and often people are chosen to get a drug based on factors that are far from random. Matching attempts to use data about subjects so that the two groups are still at least somewhat comparable. For instance, suppose we gave a drug to everyone at a particular college, so most of the participants where between 18-22 years of age. We now want to find a second group to be our control group. It wouldn't make sense to draw our control group from a set of kindergartners or from a PTA meeting. Instead we should try to find a group of people as similar to the group that took the drug as possible. Perhaps other students from a similar school could be used. Group matching is a quasi-experimental design because the groups weren't really chosen randomly, so it doesn't qualify as a true experimental design.