MULTIDIMENSIONAL ADAPTIVE PERSONALITY ASSESSMENT: A REAL-DATA DEMONSTRATION USING THE 16PF QUESTIONNAIRE
MetadataShow full item record
This thesis is a continuation of a line of research into the application of multidimensional computerized adaptive testing (MCAT) to the measurement of personality. Using archival data from the 16PF Fifth Edition, a multidimensional item response model (MIRT) was approximated allowing for nonzero item cross-loadings in order to utilize MCAT methodology, which leverages collateral information to administer items more efficiently than unidimensional CAT. Trait estimates obtained from 500 simulated MCAT administrations using actual response data were correlated with traditional CTT trait scores. Results suggest that reductions in test length of up to 50% provide estimates of personality that correlate strongly (from .70 to .86) with conventionally scored results of a full form. Previous research, possible explanations, and implications for item exposure are discussed.