Investigating Invariant Item Ordering in Intelligence Tests: Mokken Scale Analysis of KBIT-2

Özberk E. B. Ü., Özberk E. H., Uluç S., Öktem F.

International Journal of Assessment Tools in Education, vol.8, no.3, pp.714-728, 2021 (ESCI)


The Kaufman Brief Intelligence Test – Second Edition (KBIT-2) is designed to measure verbal and nonverbal abilities in a wide range of individuals from 4 years 0 months to 90 years 11 months of age. This study examines both the advantages of using Mokken Scale Analysis (MSA) in intelligence tests and the hierarchical order of the items in the KBIT-2: Turkish form by estimating the parameters of each of the three subtests by testing the dimensionality of the KBIT-2 subtests by using the Invariant Item Ordering (IIO) assumptions. 2850 people participated in the study, including children, adolescents, and adults. Participants' ages varied from 48 months (4 years 0 months) to 539 months (44 years 11 months). Automated Item Selection Procedure (AISP) was applied for the assessment of unidimensionality under three different lower bounds as 0.30, 0.40, and 0.55. The items of all three subtests formed a unidimensional scale. Backward Item Selection (BIS) procedure detected seven items in the Matrices and 17 items in the Verbal Knowledge, while six items in the Riddles subtest violated the IIO criteria. KBIT-2: Reliability values obtained using MSA analysis show that all three subtests have a high degree of internal consistency. However, care should be taken when IIO assumptions do not fit the intelligence scales in the original form.