The trials that quietly changed our lives

The independent variable (IV) was the type of glass used: toughened or regular. The dependent variables (DVs) were the level of violence and the number of injuries resulting from “glassings.” Initially, the study focused only on injuries, but it later showed a reduction in the number of attacks as well. Because participants were randomly allocated to the IV conditions, this was a true experiment, although there were still many extraneous variables.

Here are two possible examples:

• In the late 1940s, researcher Austin Bradford Hill investigated the use of streptomycin to treat tuberculosis. He randomly allocated patients to either a treatment group or a no-treatment group in order to test the effectiveness of the new antibiotic.
• In the 1950s, army doctor Thomas Chalmers randomly assigned recovering hepatitis patients to two groups: one followed the traditional prescription of two months of bed rest, while the other rested as they wished. The study tested the common belief that long periods of bed rest were necessary for recovery. Chalmers found that bed rest made no difference to the speed of recovery.

The studies described above benefited people suffering from tuberculosis and hepatitis by improving understanding of effective treatments and recovery methods.

The domestic violence experiment benefited victims of domestic violence by helping researchers identify more effective interventions and responses.

The checklist developed for use by police at immigration checkpoints benefited both suspected terrorists and the police. Suspects were treated more fairly, reflecting the principle that individuals are innocent until proven guilty, while police officers received greater cooperation from interviewees.

The three groups (3 manipulations of the IV) were giving advice, sending the perpetrator of the violence away for a cooling-off period, or arresting them.

The laws of probability still apply when participants are assigned randomly. Just as a coin flipped 10 times may land on heads 7 times, a small trial may, by chance, result in one group containing 70% of participants who are older than the average age, many more women, or more people with severe illnesses. This type of “randomization failure,” or imbalance between groups, becomes less likely as the sample size increases.

Participants in the control group, or the placebo group in drug trials, do not receive the benefit of the active treatment. Therefore, they may experience physical or psychological harm by omission. Although participants should be informed of this possibility before the trial begins, the risk of harm still raises ethical concerns. In many studies, once the experiment is complete and the treatment has been shown to be effective, researchers offer the treatment to participants in the control group as well.

Researchers may also face ethical dilemmas during a trial. For example, if evidence emerges halfway through a study that a new drug is saving lives, researchers must decide whether to continue the trial in order to obtain more reliable scientific data or to stop the study early so that the placebo group can receive the treatment.

The best-controlled studies are double-blind trials, in which neither the researchers nor the participants know who is receiving the active treatment. While this strengthens the internal validity of the findings, it can also create stress for participants, particularly those with serious illnesses, who may worry about whether they are receiving effective treatment.

Sometimes the side effects of the drugs may reveal which participants are in the experimental group. Researchers must then take care not to allow this “unblinding” to bias their observations or interpretations.

Researchers also have a responsibility to protect vulnerable populations, such as children, pregnant women, and older adults. However, excluding these groups from trials means there is often limited data about how treatments affect them, creating another important ethical dilemma.