An effect size measures how large a result is. A p-value says how surprising data are if there were no effect; the effect size says how much difference there actually was. Common versions, such as Hedges’ g, put results from different scales on one footing so studies can be combined in a meta-analysis. The American Statistical Association has stressed that statistical significance does not measure the size of an effect (ASA, 2016).
In a 2025 meta-analysis of seven LSD studies for substance use, the pooled effect was small to moderate, a Hedges’ g of 0.36 (Keighley et al., 2025). In psilocybin depression research, a 2025 analysis found that control groups in psilocybin trials improved less than control groups in antidepressant trials, which could make psilocybin look stronger than it is (Hieronymus et al., 2025).
Effect sizes are only as good as the trial behind them. A 2021 review argued that functional unblinding and expectancy likely overestimate psychedelic effect sizes (Muthukumaraswamy et al., 2021). Early, small trials often report the biggest effects, and later, larger ones shrink them.
Sources
- ASA statement on statistical significance and p-values, 2016
- Keighley et al., 2025 (PMC)
- Hieronymus et al., JAMA Network Open, 2025
- Muthukumaraswamy et al., 2021
Last checked .