Reading the research
The terms,
in everyday language.
A few words can help you understand how a study was done—and what its findings can tell you.
Randomized
Participants are assigned to study groups by chance, rather than choosing their group.
Placebo
A comparison treatment without the active ingredient being tested. It may look like the study treatment.
Blinded or open-label
Blinding means specified people do not know the treatment assignment. In an open-label study, participants and researchers know it.
Observational research
Researchers observe experiences or outcomes without assigning the treatment being studied. Other differences between people can affect the findings.
Statistical significance
A result meets a chosen statistical threshold under the analysis assumptions. That does not automatically make the effect large or useful. A result that does not meet the threshold is not proof that the treatments are identical.
Confidence interval
A range around an estimate that describes statistical uncertainty under the analysis assumptions. A narrower interval generally means a more precise estimate; it does not account for every source of bias.
Percent or percentage points?
If a measured percentage falls from 8% to 6%, that is a drop of 2 percentage points. Relative to the original 8%, it is a 25% reduction. These are different ways to describe the same change.
Group average
An average summarizes a group. Some participants can have larger changes, smaller changes, no change, or a change in the opposite direction. It is not a personal forecast.
Definitions adapted in plain language from NIH research terms and ClinicalTrials.gov study basics. Checked 7 November 2026.
Further reading: NIH: understanding study results and CDC: confidence intervals.
Three questions to take back to a profile
- Were the participants similar to the people the claim is about?
- Did researchers measure the outcome being claimed, or something else?
- What comparison and follow-up time support the conclusion?