A way to express how big an effect is, on a common scale.
The Standardised Mean Difference (SMD) is a measure of effect size that expresses the difference between two group means in units of standard deviation rather than in the original measurement units. This standardisation allows results from studies using different outcome scales (for example, different pain or disability questionnaires) to be directly compared or pooled.
When appraising a study, the SMD separates the magnitude of an effect from its statistical significance, guarding against over-interpreting a 'significant' p-value that reflects a trivially small effect, or dismissing a clinically meaningful effect that fails to reach significance in a small sample. It also protects against the error of directly comparing raw mean differences across trials that used non-identical outcome measures, which would be like comparing scores on different scales as if they meant the same thing.
SMD typically appears in meta-analyses as Cohen's d, Hedges' g, or Glass's delta, usually reported with a 95% confidence interval and often displayed in forest plots alongside individual study estimates. Readers should check which variant was used, since Hedges' g corrects for small-sample bias, and should look at the confidence interval width alongside the point estimate rather than the effect size alone.
The common thresholds of 0.2, 0.5 and 0.8 for small, moderate and large are rough conventions, not fixed biological or clinical cut-offs, and their relevance shifts depending on the condition, outcome and population studied. SMDs can also be distorted by low variability in a study's population (inflating the estimate) or by pooling heterogeneous outcome measures that are not truly comparable, so the underlying scales and study populations should be checked before interpreting a pooled SMD.
This guide was auto-drafted and is pending editorial review.