Standardized DFBeta uses what cutoff to flag influential cases?

Standardized DFBeta uses what cutoff to flag influential cases?

• A. Absolute value greater than 5.
• B. Absolute value less than -2.
• C. Absolute value greater than 2.
• D. Absolute value between -1 and 1.

Answer: (C)

Standardized DFBETA shows how much a single observation would change a regression coefficient if that observation were removed, expressed in standard-error units. The commonly used rule of thumb is to flag an observation when the magnitude of this standardized change exceeds 2. In other words, if removing the case would alter the coefficient by more than about two standard errors, that point is considered influential and deserves closer look. This threshold helps distinguish ordinary variability from cases that unduly pull the estimate. The other options don’t fit. An absolute value greater than 5 would be far less common and is not the standard benchmark here. An absolute value cannot be negative, so “less than -2” isn’t meaningful for an absolute value. An interval between -1 and 1 would indicate a very small effect, not the influential cases this statistic aims to flag.

Standardized DFBETA shows how much a single observation would change a regression coefficient if that observation were removed, expressed in standard-error units. The commonly used rule of thumb is to flag an observation when the magnitude of this standardized change exceeds 2. In other words, if removing the case would alter the coefficient by more than about two standard errors, that point is considered influential and deserves closer look. This threshold helps distinguish ordinary variability from cases that unduly pull the estimate.

The other options don’t fit. An absolute value greater than 5 would be far less common and is not the standard benchmark here. An absolute value cannot be negative, so “less than -2” isn’t meaningful for an absolute value. An interval between -1 and 1 would indicate a very small effect, not the influential cases this statistic aims to flag.