In statistics and related quantitative fields, the term mean square (abbreviated MS) denotes the average of the squares of a set of numerical values. Formally, for a collection of $n$ observations $x_1, x_2, \dots, x_n$, the mean square is
$$ \text{MS} = \frac{1}{n}\sum_{i=1}^{n} x_i^{2}. $$
The concept is employed in several contexts:
Analysis of Variance (ANOVA)
In ANOVA, mean squares are used to estimate variance components associated with different sources of variation. The total sum of squares (SS) for a source is divided by its corresponding degrees of freedom (df) to obtain the mean square:
$$ \text{MS}{\text{source}} = \frac{\text{SS}{\text{source}}}{\text{df}_{\text{source}}}. $$
These mean squares are then compared, typically via an F‑ratio, to assess the significance of effects such as treatments, blocks, or interactions.
Regression and Model Assessment
-
Mean Square Error (MSE): The average of the squared residuals (differences between observed and predicted values) in a regression model. It provides an estimate of the variance of the error term.
$$ \text{MSE} = \frac{1}{n-p}\sum_{i=1}^{n} (y_i - \hat{y}_i)^2, $$
where $p$ is the number of estimated parameters (including the intercept).
-
Mean Square Regression (MSR): In linear regression, the regression sum of squares divided by its degrees of freedom, representing the variance explained by the model.
Signal Processing and Time‑Series Analysis
Mean square values are used to quantify power or energy of signals. For a discrete‑time signal $x[n]$, the mean square value approximates its average power:
$$ P_{\text{avg}} = \frac{1}{N}\sum_{n=0}^{N-1} x[n]^2. $$
Relationship to Other Statistical Measures
- The variance of a zero‑mean variable is identical to its mean square.
- The root mean square (RMS) is the square root of the mean square, providing a measure in the original units of the data.
Historical Note
The term originates from early 20th‑century statistical methodology, where “square” referenced the squaring of deviations or observations, and “mean” indicated averaging. Its systematic use in ANOVA was popularized by Ronald A. Fisher and subsequent statisticians.
Practical Considerations
- Mean squares depend on the scale of measurement; changing units (e.g., from meters to centimeters) scales the mean square by the square of the conversion factor.
- In experimental design, reporting mean squares alongside degrees of freedom is essential for reproducibility and for the computation of F‑statistics.
See Also
- Variance
- Sum of squares
- Root mean square (RMS)
- Analysis of variance (ANOVA)
- Mean squared error (MSE)
This entry reflects commonly accepted definitions and applications of the term “mean square” in statistical literature and related scientific disciplines.