An observation is judged as an outlier if
the leverage value is greater than twice the mean
leverage value. The leverage value is
the diagonal term of the hat
matrix. In matrix notation, the hat
matrix is denoted as:
H = X(X'X)
-1X'
The leverage value indicates
whether or not the X values (indepdent variables) for the ith observation are outliers.
Specifically, it can be shown that leverage value for the ith observation is a measure of the
distance between the X values for the ith
case and the means of the X values for all n cases. The mean leverage value is the arithmetic
mean of the n leverage values or (p/n), where p is the number or variables
including the intercept term.