Selecting the most informative items in the IIP scales for personality disorders: An application of item response theory

The first goal of the present analyses was to shorten the five scales (Pilk onis, P, A., Kim, Y,, Proietti, J, M,, & Barkham, M, [1996], Journal of Per sonality Disorders, 10, 355-369) for personality disorders (PDs) developed from the Inventory of Interpersonal Problems (IIP), thereby increasing thei r attractiveness for screening purposes. The second goal was to illustrate, for more general purposes, the utility of item response theory (IRT) for s uch scale refinement. IRT analyses were performed using data collected from six different samples (N = 1149) at five sites and a two-parameter (2P) gr aded model designed for multiple response items like those on the IIP, The five most informative items from each scale were identified, based on the m agnitude of item discrimination parameters and the range and elevation of i ndividual item information functions. Preliminary analyses of the reliabili ty and validity of the short forms of the scales (totaling 25 items) suppor ted their value as alternatives to the longer forms (consisting of 47 items ), although definitive tests of their psychometric properties await crossva lidation in independent samples. Analyses of the quality receiver operating characteristics (QROC) of the long and short forms showed that both versio ns can be useful in predicting the presence versus absence of any PD diagno sis arrived at by using either a "best estimate" clinical consensus method or a structured Axis II interview.