Item response theory analyses of the Personality Assessment Inventory in samples of methadone maintenance patients and university students / by Albert Patrick Gouge.
Gouge, Albert Patrick
MetadataShow full item record
This dissertation employed a variety of item response theory models and methods to examine the psychometric function of the Personality Assessment Inventory (PAI) in a sample of opioid dependent individuals (TV = 323) receiving methadone maintenance treatment. The analyses employed both nonparametric and parametric models to examine the PAI scales and items for monotonicity, dimensionality, discrimination, difficulty, information, differential item functioning, and differential test functioning. A large sample of post-secondary students (TV = 919) were employed as a comparison group for the examination of differential item and test functioning. These analyses resulted in a potential revised version of the PAI for use in methadone maintenance populations and other substance abusing populations. Most scales and subscales were reduced in length by approximately 50% yet retained the majority of the information offered. Many scales and subscales were demonstrated to be multidimensional. Two causal factors were hypothesized with respect to this demonstration of multidimensionality. First, the widespread use of negatively scored items in the PAI likely results in the inclusion of items which are not on the same continuum as the positively worded likely due to artifactual method effects. The second major issue with respect to multidimensionality is the inclusion of symptoms of Axis I and Axis II disorders on the same scale. As well, many other items were shown to offer little in the way of discriminatory power or information. Many scales displayed items with statistically significant differential item functioning (DIF), however, only a few scales were found to have significant differential test functioning. It is suggested that this revised version of the PAI offers an improved alternative scoring method for the assessment of substance abusing populations.