Effect sizes
Hedges' g
Implemented
effect-size.hedges_gDefinition
Cohen's d with a small-sample bias correction.
Formula
g = d · (1 − 3 / (4(n₁ + n₂) − 9))
Inputs and outputs
- sample a
- sample b
Returns: EffectSizeResult
Assumptions
No assumptions beyond valid, aligned inputs of the documented types.
Limitations
No metric-specific limitations are documented yet. Interpret the value alongside the task, data, and other metrics.
Python API
import evalsuite as es
es.hedges_g(scores_a, scores_b)References
- Hedges, L. V. (1981). Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics, 6(2), 107–128.