In science and medicine, a blind study or blind experiment is one in which information about the study is withheld from the participants until the experiment ends. The purpose of blinding an experiment is reducing bias, which is a type of error. Sometimes blinding is impractical or unethical, but in many experiments it improves the validity of results. Here is a look at the types of blinding and potentials problems that arise.
Single Blind, Double Blind, and Triple Blind Studies
The three types of blinding are single blinding, double blinding, and triple blinding:
Single Blind Study
In a single blind study, the researchers and analysis team know who gets a treatment, but the experimental subjects do not. In other words, the people performing the study know what the independent variable is and how it is being tested. The subjects are unaware whether they are receiving a placebo or a treatment. They may even be unaware what, exactly, is being studied.
Example: Violin Study
For example, consider an experiment that tests whether or not violinists can tell the difference a Stradivarius violin (generally regarded as superior) and a modern violin. The researchers know the type of violin they hand to a violinist, but the musician does not (is blind). In case you’re curious, in an actual experiment performed by Claudia Fritz and Joseph Curtin, it turned out violinists actually can’t tell the instruments apart.
Double Blind Study
In a double blind study, neither the researchers nor subjects know which group receives a treatment and which gets a placebo.
Example: Drug Trial
Many drug trials are double-blinded, where neither the doctor nor patient knows whether the drug or a placebo is administered. So, who gets the drug or the placebo is randomly assigned (without the doctor knowing who gets what). The inactive ingredients, color, and size of a pill (for example) are the same whether it is the treatment or placebo.
Triple Blind Study
A triple blind study includes an additional level of blinding. So, the data analysis team or the group overseeing an experiment is blind, in addition to the researchers and subjects.
Example: Vaccine Study
Triple blind studies are common as part of the vaccine approval process. Here, the people who analyze vaccine effectiveness collate data from many test sites and are unaware of which group a participant belongs to.
Some guidelines advocate for removing terms like “single blind” and “double blind” because they do not inherently describe which party is blinded. For example, a double blind study could mean the subjects and scientists are blind or it could mean the subjects and assessors are blind. When you describe blinding in an experiment, report who is blinded and what information is concealed.
The point of blinding is minimizing bias. Subjects have expectations if they know they receive a placebo versus a treatment. And, researchers have expectations regarding the expected outcome. For example, confirmation bias occurs when an investigator favors outcomes that support pre-existing research or the scientist’s own beliefs.
Unblinding is when masked information becomes available. In experiments with humans, intentional unblinding after a study concludes is typical. This way, a subject knows whether or not they received a treatment or placebo. Unblinding after a study concludes does not introduce bias because the data has already been collected and analyzed.
However, premature unblinding also occurs. For example, a doctor reviewing bloodwork often figures out who is getting a treatment and who is getting a placebo. Similarly, patients feeling an effect from a pill or injection suspect they are in the treatment group. One safeguard against this is an active placebo. An active placebo causes side effects, so it’s harder to tell treatment and placebo groups apart just based on how a patient feels.
Although premature unblinding affects the outcome of the results, it isn’t usually reported. This is a problem because unintentional unblinding favors false positives, at least in medicine. For example, if subjects believe they are receiving treatment, they often feel better even if a therapy isn’t effective. Premature unblinding is one of the issues at the heart of the debate about whether or not antidepressants are effective. But, it applies to all blind studies.
Uses of Blind Studies
Of course, blind studies are valuable in medicine and scientific research. But, they also have other applications.
For example, in a police lineup, having an officer familiar with the suspects can influence a witness’s selection. A better option is a blind procedure, using an office who does not know a suspect’s identity. Product developers routinely use blind studies for determining consumer preference. Orchestras use blind judging for auditions. Some employers and educational institutions use blind data for application selection.
- Bello, Segun; Moustgaard, Helene; Hróbjartsson, Asbjørn (October 2014). “The risk of unblinding was infrequently and incompletely reported in 300 randomized clinical trial publications”. Journal of Clinical Epidemiology. 67 (10): 1059–1069. doi:10.1016/j.jclinepi.2014.05.007
- Daston, L. (2005). “Scientific Error and the Ethos of Belief”. Social Research. 72 (1): 18. doi:10.1353/sor.2005.0016
- MacCoun, Robert; Perlmutter, Saul (2015). “Blind analysis: Hide results to seek the truth”. Nature. 526 (7572): 187–189. doi:10.1038/526187a
- Moncrieff, Joanna; Wessely, Simon; Hardy, Rebecca (2018). “Meta-analysis of trials comparing antidepressants with active placebos”. British Journal of Psychiatry. 172 (3): 227–231. doi:10.1192/bjp.172.3.227
- Schulz, Kenneth F.; Grimes, David A. (2002). “Blinding in randomised trials: hiding who got what”. Lancet. 359 (9307): 696–700. doi:10.1016/S0140-6736(02)07816-9