How To Calculate Eta Squared In Excel

Calculate effect size for ANOVA results directly from your Excel data

Comprehensive Guide: How to Calculate Eta Squared in Excel

Eta squared (η²) is a measure of effect size that indicates the proportion of variance in the dependent variable that is explained by the independent variable in ANOVA designs. Unlike p-values which only tell you whether an effect exists, eta squared quantifies the strength of that effect.

Why Use Eta Squared Instead of Just p-values?

• Quantifies effect magnitude – Shows how much variance is explained (0 to 1)
• Comparable across studies – Unlike p-values which depend on sample size
• Required by APA 7th edition – Effect sizes must be reported with inferential statistics
• Helps with power analysis – Critical for determining sample size needs

The Eta Squared Formula

The fundamental formula for eta squared is:

η² = SS_between / SS_total

Where:

• SS_between = Sum of squares between groups
• SS_total = Total sum of squares (SS_between + SS_within)

Step-by-Step: Calculating Eta Squared in Excel

Method 1: Using ANOVA Output Directly

1. Run your ANOVA in Excel using Data → Data Analysis → Anova: Single Factor
2. Locate the key values in the ANOVA table:
   • SS Between Groups (typically labeled “Between Groups”)
   • SS Within Groups (typically labeled “Within Groups”)
   • SS Total (sum of the above two)
3. Apply the formula in a new cell:

   = [SS Between] / [SS Total]

4. Format as percentage (right-click → Format Cells → Percentage)

Method 2: Manual Calculation from Raw Data

1. Calculate group means using =AVERAGE() for each group
2. Compute grand mean using =AVERAGE() for all data
3. Calculate SS_between:

   = n₁(mean₁ – grand_mean)² + n₂(mean₂ – grand_mean)² + …

4. Calculate SS_total:

   = Σ(x – grand_mean)² for all observations

5. Compute eta squared using the formula above

Interpreting Eta Squared Values

Cohen (1988) provided these general guidelines for interpreting eta squared:

Effect Size	η² Value	Interpretation
Small	0.01 to 0.059	Explains 1-5.9% of variance
Medium	0.06 to 0.139	Explains 6-13.9% of variance
Large	≥ 0.14	Explains 14%+ of variance

Note: These are general guidelines. Some fields (like psychology) may use slightly different thresholds. Always check your specific discipline’s standards.

Common Mistakes to Avoid

• Using SS_within instead of SS_total – This gives you omega squared (ω²), not eta squared
• Ignoring assumptions – Eta squared assumes homogeneity of variance (check with Levene’s test)
• Overinterpreting small effects – Even “statistically significant” results with η² < 0.01 have negligible practical importance
• Not reporting confidence intervals – Always include CIs for effect sizes (our calculator provides these)

Eta Squared vs. Other Effect Size Measures

Measure	When to Use	Formula	Range
Eta Squared (η²)	ANOVA designs	SSbetween/SStotal	0 to 1
Partial Eta Squared (ηₚ²)	Factorial ANOVA (multiple IVs)	SSeffect/(SSeffect + SSerror)	0 to 1
Omega Squared (ω²)	Less biased estimate for population	(SSbetween – (k-1)MSwithin)/(SStotal + MSwithin)	0 to 1
Cohen’s d	t-tests (two groups)	(M₁ – M₂)/spooled	Unbounded

Advanced Considerations

Confidence Intervals for Eta Squared

Our calculator provides 95% confidence intervals using the noncentral F distribution method (Smithson, 2001). The formula involves:

1. Calculating noncentrality parameter (λ) = η²/(1-η²) × (N – k)
2. Using F distribution quantiles to find upper/lower bounds
3. Transforming back to η² scale

This method is more accurate than simple bootstrap approaches for ANOVA designs.

Handling Unbalanced Designs

When group sizes are unequal:

• Type I SS (default in Excel) gives biased η²
• Use Type II or Type III SS for more accurate results
• Consider using generalized eta squared (Olejnick & Algina, 1984)

Practical Example in Excel

Let’s walk through a concrete example with three treatment groups:

Group	n	Mean	SD
Control	30	15.2	2.1
Treatment A	30	18.7	2.3
Treatment B	30	22.4	2.0

Step 1: Run ANOVA in Excel (Data → Data Analysis → Anova: Single Factor)

Sample Output:

Source	SS	df	MS	F	p-value
Between Groups	420.13	2	210.07	42.38	1.2E-12
Within Groups	247.80	87	2.85
Total	667.93	89

Step 2: Calculate eta squared = 420.13 / 667.93 = 0.629 (62.9%)

Interpretation: This represents a very large effect size, indicating that 62.9% of the variance in the dependent variable is explained by group membership.

Reporting Eta Squared in APA Format

Proper APA reporting includes:

1. The test statistic and degrees of freedom
2. The p-value
3. The effect size with confidence interval
4. A clear interpretation

Example:

The one-way ANOVA revealed a significant effect of treatment group on outcome scores, F(2, 87) = 42.38, p < .001, η² = .63 [.52, .71], representing a very large effect.

Limitations of Eta Squared

• Biased estimator – Tends to overestimate the population effect size
• Depends on study design – Values aren’t directly comparable across different designs
• Assumes normality – Violations can inflate Type I error rates
• Not suitable for:
  • Repeated measures designs (use partial η² instead)
  • Covariance analysis (use partial η²)
  • Nonparametric tests

Alternatives When Eta Squared Isn’t Appropriate

Scenario	Recommended Measure	Calculation
Repeated measures ANOVA	Partial eta squared (ηₚ²)	SSeffect / (SSeffect + SSerror)
ANCOVA	Partial eta squared (ηₚ²)	Same as above
Nonparametric tests	Epsilon squared (ε²)	Based on rank sums
Multivariate ANOVA	Pillai’s trace or Wilks’ λ	Complex matrix operations

Frequently Asked Questions

Can eta squared be negative?

No, eta squared ranges from 0 to 1. Negative values typically indicate a calculation error (often using wrong SS values).

Why does my eta squared seem too high?

Common causes include:

• Using SS_between/SS_within instead of SS_total
• Small within-group variance (check your data for errors)
• Outliers inflating between-group differences
• Very small sample sizes can produce unstable estimates

How do I calculate eta squared for a two-way ANOVA?

For factorial designs:

1. Calculate separate η² for each main effect and interaction
2. Use SS_effect/SS_total for each
3. Consider using partial eta squared instead to focus on each effect controlling for others

What’s the difference between eta squared and R squared?

While both represent proportion of variance explained:

• R² is used in regression contexts
• η² is specifically for ANOVA designs
• R² can be negative in some cases (adjusted R²), while η² cannot

Excel Functions for Advanced Calculations

For power users, these Excel functions can help with eta squared calculations:

• =VAR.P() – Population variance (for SS_total calculations)
• =DEVSQ() – Sum of squared deviations
• =F.INV.RT() – For confidence interval calculations
• =CHISQ.INV.RT() – Useful for noncentral distributions
• =LINEST() – Can be adapted for ANOVA calculations

Authoritative Resources

For further reading on effect sizes and ANOVA:

• APA Guidelines on Effect Sizes – Official American Psychological Association recommendations
• St. Lawrence University ANOVA Guide – Comprehensive tutorial with Excel examples
• NIH Guide to Effect Sizes – National Institutes of Health publication on statistical reporting

Final Recommendations

1. Always report effect sizes with confidence intervals
2. Check assumptions (normality, homogeneity of variance) before interpreting
3. Use partial eta squared for complex designs with multiple factors
4. Consider omega squared for less biased population estimates
5. Visualize your effects – Our calculator includes a chart to help interpret the magnitude
6. Compare with meta-analyses in your field to contextualize your findings