| The following study is an Item Response Theory (IRT) analysis of the Anger Regulation and Expression Scale (ARES). A review of various applications of IRT is provided, with a focus on its recent application to measures of psychopathology and personality traits. The current study used the normative sample for the ARES and applied a unidimensional graded response model (GRM) to each of the 17 factors contained within the ARES. First, the assumptions required of a unidimensional model were tested. The GRM was determined to fit the data reasonably well, and the assumptions of unidimensionality and local independence were met. Next, IRT parameters (difficulty and discrimination) and graphical representations of the "informativeness" of the items and factor as a whole are provided. Lastly, the item with the most information in the clinical range of theta was identified in each factor, and assembled into a proposed 'IRT Short Form' of the ARES. Regression analyses indicated that both short forms reliably distinguished between non-clinical and clinical cases as well as between different clinical diagnoses. The ARES-S predicted clinical status and diagnostic category slightly better than the IRT Short Form. The implications of the findings for school psychologists as well as limitations and directions for future research are discussed. |
