Which measure assesses a point's influence on parameter estimates in a regression model?

Which measure assesses a point's influence on parameter estimates in a regression model?

• A. Features
• B. Heteroskedasticity
• C. Outliers
• D. Cook's Distance

Answer: (D)

In regression diagnostics, Cook's Distance measures how much a single observation affects the estimated coefficients when that observation is removed. It combines how far the point is from the fitted model (residual) with how influential the point is in the predictor space (leverage) to quantify the overall impact on all parameter estimates. A large Cook's Distance indicates the observation would noticeably change the coefficients if it were excluded, signaling that it’s influential and worth closer examination. Features are the predictor variables themselves, not a gauge of influence on estimates. Heteroskedasticity describes non-constant error variance, not how much a point shifts the model’s estimates. Outliers are extreme observations that may influence results, but Cook's Distance specifically quantifies the actual impact on the parameter estimates when that observation is removed.

In regression diagnostics, Cook's Distance measures how much a single observation affects the estimated coefficients when that observation is removed. It combines how far the point is from the fitted model (residual) with how influential the point is in the predictor space (leverage) to quantify the overall impact on all parameter estimates. A large Cook's Distance indicates the observation would noticeably change the coefficients if it were excluded, signaling that it’s influential and worth closer examination.

Features are the predictor variables themselves, not a gauge of influence on estimates. Heteroskedasticity describes non-constant error variance, not how much a point shifts the model’s estimates. Outliers are extreme observations that may influence results, but Cook's Distance specifically quantifies the actual impact on the parameter estimates when that observation is removed.