Index: The Book of Statistical ProofsStatistical Models ▷ Univariate normal data ▷ Simple linear regression ▷ Correlation with covariate is zero

Theorem: In simple linear regression, the residuals and the covariate are uncorrelated when estimated using ordinary least squares.

Proof: The residuals are defined as the estimated error terms

$\label{eq:slr-res} \hat{\varepsilon}_i = y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i$

where $\hat{\beta}_0$ and $\hat{\beta}_1$ are parameter estimates obtained using ordinary least squares:

$\label{eq:slr-ols} \hat{\beta}_0 = \bar{y} - \hat{\beta}_1 \bar{x} \quad \text{and} \quad \hat{\beta}_1 = \frac{s_{xy}}{s_x^2} \; .$

With that, we can calculate the inner product of the covariate and the residuals vector:

$\label{eq:slr-rescorr} \begin{split} \sum_{i=1}^n x_i \hat{\varepsilon}_i &= \sum_{i=1}^n x_i (y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i) \\ &= \sum_{i=1}^n \left( x_i y_i - \hat{\beta}_0 x_i - \hat{\beta}_1 x_i^2 \right) \\ &= \sum_{i=1}^n \left( x_i y_i - x_i (\bar{y} - \hat{\beta}_1 \bar{x}) - \hat{\beta}_1 x_i^2 \right) \\ &= \sum_{i=1}^n \left( x_i (y_i - \bar{y}) + \hat{\beta}_1 (\bar{x} x_i - x_i^2 \right) \\ &= \sum_{i=1}^n x_i y_i - \bar{y} \sum_{i=1}^n x_i - \hat{\beta}_1 \left( \sum_{i=1}^n x_i^2 - \bar{x} \sum_{i=1}^n x_i \right) \\ &= \left( \sum_{i=1}^n x_i y_i - n \bar{x} \bar{y} - n \bar{x} \bar{y} + n \bar{x} \bar{y} \right) - \hat{\beta}_1 \left( \sum_{i=1}^n x_i^2 - 2 n \bar{x} \bar{x} + n \bar{x}^2 \right) \\ &= \left( \sum_{i=1}^n x_i y_i - \bar{y} \sum_{i=1}^n x_i - \bar{x} \sum_{i=1}^n y_i + n \bar{x} \bar{y} \right) - \hat{\beta}_1 \left( \sum_{i=1}^n x_i^2 - 2 \bar{x} \sum_{i=1}^n x_i + n \bar{x}^2 \right) \\ &= \sum_{i=1}^n \left( x_i y_i - \bar{y} x_i - \bar{x} y_i + \bar{x} \bar{y} \right) - \hat{\beta}_1 \sum_{i=1}^n \left( x_i^2 - 2 \bar{x} x_i + \bar{x}^2 \right) \\ &= \sum_{i=1}^n (x_i - \bar{x})(y_i - \bar{y}) - \hat{\beta}_1 \sum_{i=1}^n (x_i - \bar{x})^2 \\ &= (n-1) s_{xy} - \frac{s_{xy}}{s_x^2} (n-1) s_x^2 \\ &= (n-1) s_{xy} - (n-1) s_{xy} \\ &= 0 \; . \end{split}$

Because an inner product of zero also implies zero correlation, this demonstrates that residuals and covariate values are uncorrelated under ordinary least squares.

Sources:

Metadata: ID: P277 | shortcut: slr-rescorr | author: JoramSoch | date: 2021-10-27, 13:07.