Science uses randomised controlled trials to test models.

In the best scientific trials, the scientists who conduct them know as little as possible.

This is one of ScienceOrNot’s Hallmarks of science. See them all here.

In short…

Randomized controlled trials are used in statistical studies of populations. The random selection process and the ability to ‘double-blind’ the participants mean that non-experimental variables can be neutralized if they cannot be controlled.

The RCT is a very beautiful technique, of wide applicability, but as with everything else there are snags. When humans have to make observations there is always the possibility of bias. To reduce this possibility a modifications has been introduced: the ‘double-blind’ randomized trial.

Archie Cochrane, British medical researcher, 1972

What are randomised controlled trials?

Randomized controlled trials (RCTs) are used to find out whether a particular human response or condition is correlated with a specified agent. In a randomized controlled trial, scientists assign subjects randomly to different groups, each of which receives a different treatment. Other than the difference in treatment, all conditions for the groups are identical. RCTs are used as a research technique in medicine, education and social sciences. They are known as randomized clinical trials when they are concerned with medical treatments.

How randomised controlled trials are conducted

In RCTs, special measures are taken to ensure subjects are assigned to groups randomly. Usually, the groups have roughly equal numbers.

Most commonly, RCTs have two groups. One group (the experimental group) receives an experimental treatment. The other (the control group) receives either the existing standard treatment (if one exists) or a placebo. A placebo is a dummy treatment which, to the subject, appears identical to the experimental treatment.

If possible, the subjects should not know which group they are assigned to. This is known as ‘concealment’ or ‘blinding’. The gold standard in RCTs is ‘double blinding’, when neither the subjects nor the experimenters know who is in which group.

Why randomized controlled trials are needed

Humans are susceptible to the placebo effect, in which they judge that their condition improves after a treatment, even a ‘dummy’ one. It occurs because aspects of their experience apart from the supposed ‘active ingredient’ influence their wellbeing. Randomized controlled trials can use a dummy treatment, or placebo, on the control group to find whether the real treatment is effective, or whether an observed phenomenon is caused by a particular stimulus. They can also be used to determine whether a new treatment is more effective than an existing one.

Bogus science tends to rely on anecdotal evidence and testimonials, neither of which eliminates the placebo effect.

A limitation of randomized controlled trials

While an RCT can reveal that a certain agent is correlated with a specific outcome, it can’t show that this is a cause-and-effect situation. Correlation does not prove causation. Evidence from other types of investigation, such as controlled experiments, is required to support a causative link between the two. Preferably, evidence of this type should be sought before an RCT is even considered.

Examples

• GlaxoSmithKline used a double-blind RCT for the Phase 3 trials of the Human Papilloma Virus vaccine. The control group received doses of hepatitis A vaccine. Each group contained over 7000 subjects and the study showed that the vaccine is effective.
• Researchers in Hong Kong used a double-blind RCT to study the effectiveness of acupuncture in improving outcomes in IVF treatment.  In this study, the pregnancy rate was significantly higher in the control group, which was treated with sham acupuncture.
• Numerous trials show little evidence that acupuncture works to reduce pain.
• RCTs have been used to assess the effectiveness of teaching reading and spelling through different systems of phonics.
• Proper RCTs show that homeopathic treatments work no better than placebos. See here, here and here.
• A 2010 study into the homeopathic treatment of rheumatoid arthritis  used a double-blind RCT. It showed that, although homeopathic consultation had some benefits, homeopathic remedies did not.
• A randomized controlled trial showed that experienced violinists are unable to tell the difference between the sound quality of old and new violins.
• Richard Saunders explains how to conduct a test on dowsing.
• RCTs have shown that intercessory prayer has no effect on the recovery of sick people.

Further reading

The Value of Carefully Controlled Studies: A Thought Experiment at The Logic of Science.

Archie Cochrane’s quote is from Effectiveness and Efficiency: random reflections on Health Services, London: Nuffield Provincial Hospitals Trust, 1972
The Clinical trial kit production image is from Fisher BioServices.

This is one of ScienceOrNot’s Hallmarks of science. See them all here.

Updated: 2013/08/29