Abstract
Local-influence analysis of perturbations of the response vector is shown to be the same as the usual regression diagnostics in the fixed-effects model. This result provides a further justification for local-influence analysis.