Effect Size in independent samples T-test
In SPSS and statistics more broadly, effect size in the context of a t-test measures the magnitude of the difference between two groups, independent of sample size. While a t-test tells you whether the difference is statistically significant, the effect size tells you how large or meaningful that difference is.
🔍 Why Effect Size Matters
- A small p-value might indicate statistical significance—but the actual difference could be tiny and unimportant.
- Effect size gives a standardized measure of how different the groups really are.
- It’s especially useful in comparing results across studies (e.g., meta-analyses).
📐 Common Effect Size Metrics in T-Tests
✅ Cohen’s d (most common for t-tests)
Used for independent samples t-tests.
Formula: d=(M1−M2)/SD_pooled
Where:
- M1 and M2 are the group means
- SD_pooled is the pooled standard deviation
Interpretation (Cohen, 1988):
- 0.2 = small effect
- 0.5 = medium effect
- 0.8 = large effect
✅ Hedges’ g
- Like Cohen’s d, but corrected for small sample bias.
✅ Eta squared (η²) and partial eta squared
- More common in ANOVA, but SPSS may show them for t-tests.
- Measures proportion of variance explained by the group difference.
📊 Where to Find It in SPSS
- When running a t-test in SPSS:
- Go to Analyze > Compare Means > Independent-Samples T Test
- Click Options → Tick “Effect Size” (if using SPSS v27+)