Leverage

in a regression or generalized linear model is a measure of the influence of a particular observation. Specifically, it is the corresponding diagonal value of the projection matrix used to fit the model.