A New IRT Approach to Test Construction and Scoring Designed to Reduce the Effects of Faking in Personality Assessment: The Generalized Graded Unfolding Model for Multi -Unidimensional Paired Comparison Responses

Responses to personality items can be influenced by motivational factors present in the testing situation. In employment settings, individuals may fake good to raise their scores and increase their chances of being hired; in judicial settings, individuals may fake bad to lower their scores and reduce the likelihood of being judged mentally competent. Historically, researchers have tried to reduce the untoward effects of faking by using post hoc score adjustments, based on social desirability scores, but these approaches have had little salutary effect on validity and utility. Consequently, this research focuses on developing personality items and tests that are fake-resistant. Paired comparison responses to personality items, consisting of pairs of statements on different dimensions, which are similar in social desirability, are scored using a new multi-unidimensional item response theory model, called the GGUM-MU. Equations for the GGUM-MU item response and information functions are derived, a statistical method for examining model-data fit is illustrated, and the results of simulation studies to investigate the accuracy of a Bayes modal latent trait estimation procedure for one- and two-dimensional tests of various lengths are presented. The potential application of the GGUM-MU approach to personality test construction in organizational settings is also addressed.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Psychology

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Stark, Stephen Edward

Contributors dc:contributor

• Drasgow, Fritz

• Psychology, Personality

Language dc:language

eng

Identifier

(MiAaPQ)AAI3044232

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/82023