Advertisement

Effect size calculator

Describe the size and direction of an observed association. Choose a measure that matches the design and standardisation used in your study.

Calculator

Cohen d
1.000000
Hedges g (exact correction)
0.957646
Glass delta, group 2 reference
1.000000
Pooled sample SD
2.000000

Advertisement

How this calculator works

Cohen d standardises the group-1 minus group-2 mean difference by a pooled sample SD. Hedges g applies the exact gamma-function correction for small samples. Glass delta uses group 2 as the reference SD. The t conversion assumes a pooled-variance independent-groups t test; it is not a paired or Welch conversion. Correlation mode returns r and r² without forcing them into a mean-difference measure.

Formula

s_p² = [(n1-1)s1²+(n2-1)s2²]/(n1+n2-2); d = (mean1-mean2)/s_p; g = J(df)d; J(df) = Γ(df/2)/(√(df/2)Γ((df-1)/2)); d = t√(1/n1+1/n2).

Worked example

With means 12 and 10, both SDs 2 and both sample sizes 10, pooled SD is 2, d is 1 and exact Hedges g is about 0.9576464.

Advertisement

Frequently asked questions

Does a large effect prove causality?

No. Study design, confounding and uncertainty still matter.

What does the sign mean?

For group summaries, positive means group 1 has the higher mean. Reversing group order reverses the sign.

Can I use universal small, medium and large labels?

Such labels are rough conventions. Interpret the size against real outcomes and relevant prior studies.

Sources

Advertisement