Glossary · Methodology

Effect size

In short

Effect size describes how large a difference is — not just whether it exists. It is often given as a standardized mean difference so that studies with different measurement scales can be compared.

An effect can be given in minutes, kilograms or points on a scale. The standardized mean difference converts it into multiples of the natural spread. As a rough rule of thumb, 0.2 counts as small, 0.5 as medium and 0.8 as large.

The rule of thumb does not replace the question of whether a difference is noticeable in everyday life. Falling asleep seven minutes faster can be statistically well established and still barely noticeable for the individual. Conversely, a small effect on a hard endpoint can mean a lot.

Only the size of the effect tells you whether it counts.

Where this comes up in the fact checks

HMB – In patients with muscle wasting, muscle strength rose by a standardized mean difference of 0.31 — a small effect.
Melatonin – In ordinary insomnia about seven minutes faster sleep onset, in delayed sleep phase syndrome 38.8 minutes.

Sources

Sullivan 2012, J Grad Med Educ – using effect size, or why the P value is not enough (PMID 23997866)

All terms in the glossary

Information only, not medical advice. The BK-Score is a subjective assessment according to disclosed rules – not a scientific evaluation. Scoring rules