Which term describes the total unexplained variability across observations, as measured by squared deviations from model predictions?

Which term describes the total unexplained variability across observations, as measured by squared deviations from model predictions?

• A. Residual sum of squares
• B. Residuals
• C. Reliability
• D. Reverse Helmert contrast

Answer: (A)

In regression and related models, the observed variability in Y that the model cannot explain is captured by the residual sum of squares. This quantity sums the squared differences between each observed value and its predicted value from the model, giving a single measure of the total unexplained variability: RSS = Σ(Yi − Ŷi)². It’s a key piece in assessing model fit and in computing R-squared, since R-squared = 1 − RSS/TSS. The residuals themselves are the individual deviations (Yi − Ŷi), not their total squared sum. Reliability refers to consistency or repeatability, and a reverse Helmert contrast is a specific way of coding categorical variables for comparisons—neither describes the total unexplained squared deviation from the model.

In regression and related models, the observed variability in Y that the model cannot explain is captured by the residual sum of squares. This quantity sums the squared differences between each observed value and its predicted value from the model, giving a single measure of the total unexplained variability: RSS = Σ(Yi − Ŷi)². It’s a key piece in assessing model fit and in computing R-squared, since R-squared = 1 − RSS/TSS. The residuals themselves are the individual deviations (Yi − Ŷi), not their total squared sum. Reliability refers to consistency or repeatability, and a reverse Helmert contrast is a specific way of coding categorical variables for comparisons—neither describes the total unexplained squared deviation from the model.