Controlled and Uncontrolled Studies

Some terminology:

In a controlled study, researchers are able to determine which of their subjects receive the factor that is being tested for having a causal influence upon another factor. The alleged causal factor is sometimes called the independent variable, while the alleged effect is called the dependent variable.

Those given the alleged causal factor make up the experimental group, while those who don’t receive such treatment belong in the control group. Ideally, both groups will be balanced with respect to the subjects’ various other characteristics. Often this is achieved simply by randomly assigning subjects to experimental or control groups.

In a blind study, the subjects do not know whether they are in the control group or not. In a double blind study, the researchers gathering the data will not know which group a subject belongs to. Blinding is meant to mitigate placebo effects – possible confounding of the study by the subjects’ or researchers’ knowledge.

In uncontrolled or observational studies, researchers have no such control over whether their subjects receive the treatment being investigated. Uncontrolled studies thus are subject to worries about confounding variables.

An example:

Earlier this month, professor Yasmin Hurd of the Mount Sinai School of Medicine released a study showing that rats exposed to the main ingredient in marijuana during their adolescence showed a greater sensitivity to heroin as adults. The wire lit up with articles announcing confirmation for the "gateway theory"—the claim that marijuana use leads to harder drugs.

It's a theory that has long seemed to make intuitive sense, but remained unproven. The federal government's last National Survey on Drug Use and Health, conducted in 2004, counted about 97 million Americans who have tried marijuana, compared to 3 million who have tried heroin (166,000 had used it in the previous month). That's not much of a rush through the gateway. And a number of studies have demonstrated that your chances of becoming an addict are higher if addiction runs in your family, or if heroin is readily available in your community, or if you're a risk-taker. These factors can account for the total number of heroin addicts, which could make the gateway theory superfluous. (Slate Magazine, July 23, 2006)

This passage cites two studies, one is controlled, the other is not. Can you identify which?

It then warns us of potential confounding factors which might also explain the observed correlation. What are these?