An item response theory analysis of the Self-as-Context Scale

The major purpose of this project was to further evaluate the psychometric properties of the Self-as-Context Scale (SACS) and its 10 items by conducting an item response theory (IRT) analysis. Prior to doing so we conducted exploratory and confirmatory factor analyses with an aggregate college student sample (N = 1842) that confirmed an acceptable fit for a bidimensional model after dropping an item from the transcending subscale. The General Response Model we chose for the IRT analysis indicated strong discriminating power for both sets of remaining subscale items in their ability to adequately differentiate varying levels of their respective dimensional constructs. An evaluation of difficulty parameters suggested both sets of items covered an adequate span of latent traits, but that each might be improved, albeit in differing ways. For centering items, this would entail the elimination of the neutral response option from the Likert-type scale used in rating them, while increasing the difficulty level of transcending items through some rewording could be considered. Implications as well as limitations of the findings and ways of addressing them in the service of further evaluating and improving the quality of SACS as an assessment tool in clinical research and practice are discussed.