Which diagnostic statistic is commonly used to identify influential observations by combining residual size and leverage?

• A. Cook's Distance
• B. Ridge
• C. Multicollinearity
• D. Heteroskedasticity

Answer: (A)

The idea being tested is how to spot observations that could unduly affect regression results by looking at both how far they fall from the fitted values and how much leverage their predictor values grant them. Cook's distance does this in one summary metric by assessing how much the estimated coefficients would change if that single observation were removed. If an observation has a large residual and high leverage, its Cook's distance becomes large, signaling potential influence on the model fit. This makes it a focused diagnostic for identifying influential data points that deserve closer inspection.

Other concepts describe different issues: a technique that regularizes coefficients to prevent overfitting, a situation where predictor variables are highly correlated, or a pattern where residual variance changes with the level of a predictor. None of these measure the influence of individual observations in the combined way that Cook's distance does.