Mean Square (MS)

Introduction: MS

The Mean Square (MS) is a statistical measure used in analysis of variance (ANOVA) and regression. It represents the average variability, calculated by dividing the sum of squares (SS) by its corresponding degrees of freedom (df). Mean squares are essential for hypothesis testing, particularly when evaluating differences between groups or the explanatory power of regression models.

Background

In ANOVA, the total variability in data is partitioned into components (e.g., between-group and within-group variation). By standardising sums of squares using degrees of freedom, mean squares provide comparable measures of variation. These values form the basis of the F-statistic, which tests whether observed differences are statistically significant.

Formula

Where:

• SS (Sum of S s   aquares) = a measure of total variation.
• df (Degrees of Freedom) = the number of independent values that can vary.

Types of Mean Squares

• Mean Square Between (MSB): Variation between group means.

• Mean Square Within (MSW): Variation within groups (residual error).

• Mean Square Regression (MSR): Variation explained by the regression model.
• Mean Square E srror (MSE): Unexplained variation or residuals.

Applications / Examples

• ANOVA: The ratio of mean squares (e.g., MSB ÷ MSW) produces the F-statistic, used to test whether group means differ significantly.
• Regression: Mean squares determine how much variation in the dependent variable is explained by the model versus random error.

Worked Example (One-Way ANOVA):

• SSB = 120, dfB = 3
  
• SSW = 80,  dfW = 16
  

The F-statistic is:

Relevance / Impact

Mean squares standardise sums of squares by accounting for degrees of freedom, making them directly comparable. They are central to ANOVA and regression analysis, enabling researchers to test hypotheses, evaluate model fit, and interpret whether observed patterns are statistically meaningful.

See also

• Analysis of Variance (ANOVA)
• Regression analysis
• Hypothesis testing
• Degrees of Freedom (df)