Standardized DFFit is used to identify ...

• A. To identify influential observations in regression.
• B. To test multicollinearity among predictors.
• C. To assess residual normality.
• D. To estimate model fit via R-squared.

Answer: (B)

Standardized DFFit flags observations that unduly affect the fitted values in a regression model. It evaluates how much the predicted value for each case would change if that single observation were left out, then scales that change so you can compare across cases on a common footing. When a case has a large absolute standardized DFFit, it means that removing that point would noticeably alter the regression line and the predicted values, indicating that the observation is influential. Such points can distort coefficient estimates and overall predictions, so identifying them helps you decide whether to investigate data quality, model specification, or whether to perform robust analyses that are less sensitive to outliers.

This diagnostic is about influence on predicted values, not about whether predictors are related (multicollinearity is typically checked with VIF or tolerance), not about whether residuals are normally distributed (assessed with normality tests or Q-Q plots), and not about overall model fit (R-squared).