According to the rule for influential cases, deleting a case will damage the precision of some model parameters if the CVR exceeds which threshold?

• A. 1 + [3(k + 1)/n]
• B. 1 − [3(k + 1)/n]
• C. 2
• D. 0

Answer: (A)

When judging whether a data point is influential, the idea is that removing it should not drastically reduce the precision of the estimated coefficients. The CVR, or change in variance ratio, measures how much the precision (the variance of the estimates) would change if that case were deleted. A high CVR means the case is having a big impact on precision.

The rule of thumb sets the threshold for CVR at 1 plus three times (k + 1) divided by n, where k is the number of predictors and n is the sample size. The k + 1 accounts for the intercept. This threshold grows with model complexity (more predictors) and shrinks with larger samples, reflecting that more data makes estimates more stable and that adding predictors can make some points more influential.

So, a case flags as damaging to precision when its CVR exceeds 1 + 3(k + 1)/n. Other fixed numbers don’t adjust for model size or data amount, which is why they are not the right threshold.