Abstract
Test Description
The Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II; Wechsler, 2011) is a brief intelligence test designed for individuals aged 6 through 90 years. It is a revision of the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999). During revision, there were three goals: enhancing the link between the Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV; Wechsler, 2003) and the Wechsler Adult Intelligence Scale, Fourth Edition (WAIS-IV; Wechsler, 2008a), increasing user-friendliness, and improving psychometric properties. Items from the WASI were dropped, modified, or retained after comparing them to items and scoring criteria on the WISC-IV and WAIS-IV. Furthermore, administration rules (e.g., reversal and discontinue rules) were modified in order to increase the consistency with other Wechsler instruments. Norms were updated, subtest floors and ceilings were extended, and reliability and validity evidence was generated.
Like the WASI, the WASI-II is composed of four subtests designed as alternate forms of the WISC-IV and WAIS-IV subtests. The Block Design subtest requires the examinee to complete a series of two-color patterns using blocks. During the Vocabulary subtest, the examinee provides the name of pictures or defines words presented orally and in print. Next, the Matrix Reasoning subtest allows the examinee to view matrices and apply inductive reasoning to select the correct response. Finally, the examinee either matches pictures or describes how common items or concepts are alike on the Similarities subtest.
Administration, Scoring, and Interpretation
Administration of all four subtests takes approximately 30 minutes. Materials needed for administration are few: test record, stimulus book, blocks, and examiner’s manual. WASI subtest instructions were simplified and shortened for the WASI-II.
The WASI-II yields two Full Scale IQs (FSIQs). The FSIQ-4 stems from scores from all four subtests, and the FSIQ-2 stems from scores from only Vocabulary and Matrix Reasoning. In addition, the WASI-II produces a Verbal Comprehension Index (VCI) stemming from scores from Vocabulary and Similarities and a Perceptual Reasoning Index (PRI) composing of scores from Block Design and Matrix Reasoning.
The FSIQ-4, FSIQ-2, VCI, and PRI yield standardized scores (M = 100, SD = 15). Confidence intervals, based on calculation of the standard error of estimation and centered around the estimated true score, are also provided. A range of expected FSIQs on the WISC-IV or WAIS-IV can also be obtained based on the FSIQ-4. The four subtests yield T scores (M = 50, SD = 10) and age equivalent scores. Tables are also available to calculate the statistical significance and base rate of VCI and PRI discrepancies. There is no scoring or report writing software.
Technical Adequacy
Standardization and Norming
The WASI-II norming sample included 2,300 individuals aged 6 to 90 years and was stratified using the 2008 United States census data. Data were collected between January 2010 and May 2011 across approximately 35 states. Although the WASI-II includes a nationally representative norming sample appropriate for general use in the United States, at this time, no Canadian norms are available (cf. Wechsler, 2004, 2008b). Norms table blocks range from 4 months (ages 6:0 to 16:11), 3 years (ages 17 to 19), 5 years (ages 20 to 84), and 6 years (ages 85 to 90). Although 100 norm sample participants are associated with each norm table block after age 16, we estimate that there are approximately 33 participants associated with each norm block from ages 6:0 to 16:11.
Scale/Item Characteristics
There were no subtest floor or ceiling violations throughout the norms table blocks. Each subtest raw score of 1 produced a T score at least two standard deviations below the mean, and each maximum subtest raw score produced a T score at least two standard deviations above the mean. The FSIQ-4 can range from 40 to 160, whereas the FSIQ-2, VCI, and PRI can range from 45 to 160. Thus, the WAIS-II is capable of measuring intelligence at both extremes of functioning.
Reliability
Corrected split-half reliability coefficients for all composites met or exceeded .90 across 1-year samples of children ages 6 through 16. The same pattern was true for the adult samples. Only the FSIQ-2 did not meet the .90 criterion at ages 17 to 19, yet its coefficient was .89. The FSIQ-4 demonstrated an average coefficient of .96 across the child sample and .97 across the adult sample. Corrected test–retest reliability coefficients across 12 to 88 days were above .90 for the FSIQ-4, FSIQ-2, and VCI across age groups. The PRI demonstrated corrected test–retest reliability coefficients below this criterion; they ranged from .86 to .87 in three of the four age groups. Inter-scorer agreement coefficients for Block Design and Matrix Reasoning ranged from .98 to .99, and for Vocabulary and Similarities, they ranged from .94 to .95. Overall, reliability evidence for the FSIQ-4 and FSIQ-2 is particularly impressive, considering how few subtests contribute to these composites.
Validity
Validity evidence based on content is supported by results from item tryouts, item scaling analysis, and item-bias analysis, but no detailed results of these analyses were provided in the test manual. As evidence of validity based on internal structure, intercorrelations between WASI-II subtests and intercorrelations between composite scores were presented, but this evidence is not particularly interesting compared to that of tests including a greater variety of (and more) subtests and composites. Perhaps most noteworthy, however, was the correlation of .94 between the FSIQ-4 and the FSIQ-2 across samples.
An array of convergent and discriminant validity evidence (stemming from correlations between and cross-battery factor analysis of the WASI-II and other intelligence test scores) and clinical group comparisons were offered. Comparisons across tests produced predictable results—both in terms of the magnitude of correlations and mean differences in norm-based scores consistent with the Flynn effect (Flynn, 1984). Of note, the FSIQ-4 correlated .91 with the WASI FSIQ-4 and WISC-IV FSIQ and .92 with the WAIS-IV FSIQ; correlation values were lower for the FSIQ-2. Correlations between the WASI-II and the Kaufman Brief Intelligence Test, Second Edition (KBIT-II; Kaufman & Kaufman, 2004) and the Wechsler Fundamentals: Academic Skills (WFAS; Wechsler, 2008c) increased the quality of this validity evidence. It was reported that the relations between the WASI-II scores and the KBIT-II scores were strong to very strong and that the relations between the WASI-II scores and the WFAS scores were moderate in magnitude, but the actual correlation values were not presented. In addition, no correlations were presented between WASI-II scores and other full-length intelligence or achievement tests. In particular, additional research is needed to examine the relations between the WASI-II scores and achievement outcomes.
When submitted to an exploratory factor analysis, WASI-II subtest scores and scores from the WISC-IV and WAIS-IV—other than those shared with the WASI-II—produced factor loadings supporting a four-factor model like those associated with these full-length tests. A series of confirmatory factor analytic models of the WASI-II subtest scores were tested across age-based subsamples from the norming sample. All contrasted a one-factor model (consistent with the FSIQ-4) with a two-factor model (consistent with the VCI and PRI subtest groupings). The two-factor model was significantly better fitting in every analysis, but (a) the correlations between the two factors (which might suggest the presence of a second-order g factor) were not reported and (b) additional subtests were not considered that would allow for testing of alternate models suggesting the separation of the WASI-II Block Design and Matrix Reasoning subtests as measures of Visual Processing and Fluid Reasoning, respectively (Benson, Hulac, & Kranzler, 2010; Keith, Goldenring Fine, Taub, Reynolds, & Kranzler, 2006). Although there is frequent reference to the g loadings of the WASI-II subtests (correlations with the general factor of intelligence) and suggestions that subtests were selected for inclusion based, in part, on these properties, no subtest g loadings are reported.
Clinical group studies including relatively small numbers of participants (ns = 21 to 54) provided some validity evidence supporting the use of the WASI-II with those with intellectual disability, intellectual giftedness, attention-deficit/hyperactivity disorder, learning disorders in reading or mathematics, and traumatic brain injuries. WASI-II scores from these samples were contrasted with those from children and adults from the WASI-II norming sample matched on key demographic variables. Analyses were basic and yielded only descriptive statistics and t-test results. More advanced studies of classification accuracy and profile similarity were not reported.
Commentary and Recommendations
It is clear that the developers of the WASI-II accomplished many of their goals in increasing administration ease and improving the psychometric properties. For those accustomed to administering the WASI, we found several beneficial differences of administering the WASI-II. The developers of the WASI-II added items to subtests to increase floors and ceilings while concomitantly reducing the number of total items; it seems that many mid-range items were omitted. The reduced numbers of items along with more stringent discontinue rules result in shorter testing sessions. Furthermore, scoring VCI subtest items is easier due to the inclusion of additional example responses. Considering standards for intelligence test norming and measurement, the updated WASI-II norms were well developed, nationally representative, and sufficiently large. Consistent with design goals, the WAIS-II subtests produce no floor or ceiling violations. In terms of reliability and validity evidence, the FSIQ-4 is a superior measure compared to the FSIQ-2. In general, we believe that the WASI-II is much improved compared to its predecessor and that it offers considerable benefits to users.
The developers of the WASI-II repeatedly conveyed that test users should restrict their use of the WASI-II to only a narrow set of purposes. They stated that “the main uses of the WASI-II are retesting and screening for a quick estimate of cognitive functioning” (Wechsler, 2011, p. 12), that it “is not intended to substitute for the FSIQ obtained on more comprehensive measures of intelligence” ( p. 9), and that “the WASI-II FSIQ should not be used for legal, judicial, or quasi-legal purposes” ( p. 10). However, the strong psychometric properties of the WASI-II composites in general and the FSIQ-4 in particular make us question the benefits of administering a full-length intelligence test when only an IQ is required as part of a comprehensive, multimethod psychological assessment. We encourage both clear thinking about the practical benefits of employing brief intelligence tests in lieu of full-length intelligence tests and also additional research to address their viability in diagnostic and eligibility decision making.