The leverage of an observation is a measure of how far the explanatory variables recorder for that observation are from the other such points. It does not consider the value of the target variable for that observation at all.
Cooks distance and dffits on the other hand consider them simultaneously.
So, if an observation hasvery typical explanatory variables recorded, but the dependent variable is unusually large or unusually small, the leverage would be small, but cooks distance and dffits would be high.
For example, if we are regressing Weight on Height as explanatory variable, and you have average Height, then you'd be a low leverage point. But if you're unusually thin for your height and weigh very less, you'd be what we call an "outlier". So you'd have high cooks distance and dffits values.
Get Answers For Free
Most questions answered within 1 hours.