The PANAS structure revisited: on the validity of a bifactor model in community and forensic samples.