Relations between the Kaufman Brief Intelligence Test (KBIT) and the Wechsler Intelligence Scale for Children - Third edition (WISC-III) in Children with Reading Disability

Concurrent validity of the Kaufman Brief Intelligence Test (K-BIT) with the Wechsler Intelligence Scale for Children-Third Edition (WISC-III) was evaluated, as well as the K-BIT's accuracy as a predictor of WISC-III scores, in a sample of young children with reading disabilities. The two measures were administered to 65 children from Atlanta, Boston, and Toronto who ranged from 6-5 to 7-11 years of age at testing. Correlations between the verbal, nonverbal, and composite scales of the K-BIT and WISC-III were .60, .48, and .63, respectively. Mean K-BIT scores ranged from 1.2 to 5.0 points higher than the corresponding WISC-III scores. Standard errors of estimation ranged from 10.0 to 12.3 points. In individual cases, KBIT scores can underestimate or overestimate WISC-III scores by as much as 25 points. Results suggest caution against using the K-BIT exclusively for placement and diagnostic purposes with young children with reading disabilities if IQ scores are required. Intelligence assessment currently plays a central role in the diagnosis and placement of children with developmental reading disability (RD). Given that individual intelligence assessments require 1 to 2 hours of testing time, shorter screening instruments that provide an estimate of individual intellectual abilities are being used with some regularity by professionals, schools, and clinical researchers as alternative measures of intellectual development. To be useful, however, screening measures must provide a valid estimate of intelligence scores. It is essential to show their concurrent validity with frequently used, more comprehensive measures of childhood intellectual abilities, such as the Wechsler Intelligence Scale for Children-Revised (WISC-R; Wechsler, 1974) and the current instrument, the Wechsler Intelligence Scale for Children Third Edition (WISC-III; Wechsler, 1991). Evaluation of the concurrent validity of screening instruments is particularly critical for children with RD because any overestimate or underestimate of IQ scores can affect diagnosis and eligibility decisions. One screening measure, the Kaufman Brief Intelligence Test (K-BIT; Kaufman & Kaufman, 1990), has proven to be a valid estimate of intelligence (as measured by the scales) in a wide range of groups, including nondisabled samples of children and adolescents (Kaufman & Kaufman, 1990), students referred for multidisciplinary evaluation (Canivez, 1995), students referred for poor academic progress (Prewett, 1992a, 1995), incarcerated juvenile delinquents (Prewett, 1992b), children with traumatic brain injury (Donders, 1995), and youth in psychiatric hospitals (Javorsky, 1993). In addition, Slate, Graham, and Bower (1996) found statistically significant correlations between the K-BIT andWISC-R in a combined sample of children with mental retardation and children with specific learning disability (data concerning the types of learning disabilities included in the sample were not provided). In general, these studies found significant correlations between K-BIT IQ Composite scores and Full Scale IQ (FSIQ) scores (range = .64-.87), as well as between K-BIT verbal index scores (i.e., Vocabulary scores) and Verbal IQ (VIQ) scores (range = .67-.83). Correlations between nonverbal scores (i.e., K-BIT Matrices and Performance IQ [PIQ]) were somewhat lower (range = .50-.74)--and, in one instance, nonsignificant (.29; Prewett, 1992b). In addition, these studies have generally found mean K-BIT IQ Composite scores to be slightly lower than mean WISC-R FSIQ scores (range = 1.0-6.2 points) and slightly higher than mean WISC-III FSIQ scores (range = 1.6-7.7). This is consistent with the general finding that WISC-R scores tend to be higher than WISC-III scores. Given that the relationship between the K-BIT and the Wechsler scales may vary across populations (Prewett, 1992a), caution must be exercised when generalizing the aforementioned findings to other groups of children, particularly groups commonly referred for testing, such as children with RD. Even in the most conservative epidemiological estimates, RD is identified in 3% to 6% of school-age children (depending on its definition) and is widely considered the most common type of developmental learning disability (Hynd & Cohen, 1983; Stanovich, 1986). Assuming that the K-BIT is a valid estimate of intelligence in an RD population is problematic, however, because the relation of the K-BIT to more comprehensive measures of intelligence has never been examined in a sample composed exclusively of children with RD. Concerns about widespread K-BIT use also must be raised when attempts are made to generalize previous findings to groups composed of children from different ethnic backgrounds, socioeconomic levels, and/or geographic regions. Although many studies examining the relations between the K-BIT and the Wechsler scales have included racially diverse samples (Canivez, 1995; Prewett, 1992a, 1992b, 1995), none of the previous studies reported the socioeconomic characteristics of their sample or included participants from more than one geographical location. Given the age-dependent specifications of K-BIT administration, the generalizability of previous results from samples with broad age ranges (i.e., 7 to 16 years of age) to samples of younger children is also problematic. The K-BIT is composed of three subtests, but the Definitions subtest is administered only to individuals 8 years of age and older. Consequently, because previous studies used samples in which this particular portion of the K-BIT was usually administered (i.e., most participants were 8 or older), generalizability to younger children (i.e., under 8 years) whose Expressive Vocabulary score is the only contribution to the K-BIT verbal index score remains uncertain. Obviously, the relation of the K-BIT verbal index scores to scores on more comprehensive measures of verbal intelligence might be significantly influenced by these differential administration requirements. Concurrent validity of the K-BIT in younger children, however, is important, given that intervention at early stages of RD has the most significant impact on long-term treatment outcome (Blachman, 1997; Foorman, Francis, S. Shaywitz, B. Shaywitz, & Fletcher, 1997), and that most children are referred for RD evaluations after first grade but before fourth grade (i.e., between the ages of 6 and 9). In addition to validity coefficients (i.e., correlations) and mean differences, the standard error of estimation is another critical factor to examine when assessing the validity of the K-BIT as an estimate of intellectual abilities. All available studies, however, failed to report standard errors of estimation when using the K-BIT scores to predict scores on the Wechsler scales, with the exception of Donders (1995). Donders found high standard errors of estimation between Vocabulary and VIQ (6.87), Matrices and PIQ (10.39), and IQ Composite and FSIQ (7.65), despite moderate to high correlations between the K-BIT scores and their respective WISC-III IQ scores (range = .52-.79). The present study examined the relation between K-BIT and WISC-III scores in a sample of young children with RD, all of whom were below 8 years of age at the time of K-BIT administration. Children were also included from different geographical regions, different socioeconomic levels, and different racial groups, to better assess the accuracy of generalization.