← All months

August 2025 1 material

How to calculate reliability

Every study gets a score from 1 to 10 for the reliability of its conclusions. The score depends on:

  • Publication type. Systematic reviews and meta-analyses of RCTs score highest. Individual RCTs and cohort studies come lower. Preprints (bioRxiv, medRxiv) score even lower due to a lack of peer review.
  • Sample size. A study with 10,000+ participants is more reliable than a pilot study with 14.
  • Journal and peer review. Publication in an impact-factor journal is not the same as a preprint.
  • Methodology. Control groups, randomisation, blinding, duration of follow-up.
  • Conflict of interest and funding. Declared and transparent — a plus.

The score is generated by AI from publication metadata. It is a Guide, not a verdict — even a 6/10 can contain an important result, and 9/10 doesn't exempt you from reading critically. Every source is signed: click "Read research →" to verify for yourself.

PubMed
The Tohoku Journal of Experimental Medicine
14 August
Reliability: 7/10 Reliability is the assessment of how confident we are in the study's conclusion. It accounts for publication type (RCT > cohort > case series > preprint), sample size, journal, and methodology. 9–10: strong evidence. 7–8: confident with caveats. 5–6: preliminary, needs confirmation. The score is generated by AI from publication metadata.

A Traditional Japanese Samurai Movement Rei-ho as Knee Extension Strength Training: A Randomised Controlled Study.

Ogasawara, A, Sato, A, Nagatomi, R

English summary coming soon.

Strength training Lower limbs Functional movements Randomised controlled trial
Read research PDF
DOI: 10.1620/tjem.2025.J099