The Kaufman Brief Intelligence Test (K-BIT - Kaufman & Kaufman, 1990) was designed to provide an estimate of the level of cognitive functioning and serve as a quick screening instrument (taking approximately 30 minutes to administer) rather than a diagnostic tool. It covers the age range of 4 to 90 years of age and involves one verbal subtest of 45 Expressive Vocabulary items and 37 Definitions, and one nonverbal subtest of 48 Matrices. There are three scores yielded from the K-BIT (Verbal, Matrices and IQ Composite) which are expressed in deviation IQ units. While the K-BIT often achieves higher reliability coefficients than other scales, a perusal of the manual accompanying the test provides evidence that in terms of reliability, validity and norms, the test itself is a sound measure.

In the manual accompanying the test, Kaufman & Kaufman (1990) detail the various techniques and the results obtained in assessing whether the K-BIT is psychometrically adequate. 2022 participants in the testing plan, aged 4 to 90 years, took part in examinations involving measures such as twenty validity studies and a range of reliability analyses in order to provide a standardized example of the K-BIT's effectiveness. This standardized sample provides an excellent example of a representative group, or norm, to which further administrations of the test can be compared.

Reliability, broadly defined as the consistency of test scores, was demonstrated by a test-retest reliability study and an application of the Standard Error of Measurement (SEM) to assess the error surrounding perceived "true" scores (mean scores repeatedly obtained if no outside factors are complicating measurement). These studies obtained a reasonable correlation coefficient, which expresses the degree of correspondence (the relationship) between two sets of scores, the best result being a value of +1.00. While obtaining this result is advantageous, it is not likely to occur, and even if it does is often assumed to be erroneous.

In examining the results of the test-retest analysis for the K-BIT , a coefficient for each of the three scores in Vocabulary, Matrices and IQ Composite is obtained for each age group. Reliability increases as age increases, the age groups being determined as 5-12 (N=53), 13-19 (N=60), 20-54 (N=69), and 55-89 (N=50). In all instances, the reliability coeffecient found for the K-BIT IQ Composite exceeded .90, which provides persuasive evidence that scores obtained from the K-BIT are remarkably reliable. This reliability is supported by an examination of the two test subcomponents, where we also find high values for the coefficient, not one of them dropping below .80. The only stipulation made by Kaufman & Kaufman (1990) concerning these results is that mean values between the first and second testing differed by 3 standard score points across all age groups, a phenomenon which they correctly assume to be due practice effect across test administrations.

The reliability of the subtests themselves are demonstrated by intercorrelations between the scores which provided evidence that, while each subtest provided a unique measure of intelligence, their combination for the IQ Composite is vindicated (4-7 achieved an intercorrelation of .44, while 8-19 and 20-90 achieved .60 and .72 respectively).

Standard Error of Measurement (SEM) provides a further dimension to the analysis of reliability in determining the confidence with which we can assert that our score fell within a particular range, usually interpreted as being a 68 percent chance that the measurement fell within this band of scores. Therefore, the smaller the SEM the better, as we can be more confident in the obtained score. As the K-BIT is extremely reliable, this is definitely the case, the SEM for the adult (20-90) IQ Composite being only 2.7, and 4.2 for children and adolescents (4-19), which averages at 3.9 for the entire age range (4.1 for Vocabulary and 5.4 for Matrices). Overall, the standardized results in the K-BIT manual demonstrate excellent reliability, a desirable outcome for any norm group.

The validity of the K-BIT is also impressive, which means we are justified in making inferences based on the results obtained from any administration. Not only is the K-BIT a valid measure of intelligence, demonstrated by its construct validity (in comparison with the WAIS-R, K-ABC and WISC-R tests), but it also provides excellent concurrent validity in comparison with achievement tests (the Kaufman Test of Educational Achievement (K-TEA), in both Brief and Comprehensive forms). The only low correlations for concurrent validity seem to be with the Test of Nonverbal Intelligence (TONI) where an IQ Composite value of .23 was obtained. This can be explained when we acknowledge that the K-BIT has a great emphasis on vocabulary in measuring intelligence (the Verbal subtest) which would act to the detriment of the correlation. Indeed, a negative correlation of -.04 is found for the Verbal subtest in comparison with the TONI.

Despite the appealing results concerning the reliability and validity of the K-BIT norm, and its comparative nature to other intelligence tests, it is not recommended as a diagnostic tool. The emphasis for the use of the K-BIT is as a screening test, and this recommendation is most certainly justified. The primary reason for this suggestion is that the K-BIT is only one source of data, and one would be making a grave mistake as a scientist performing an empirical evaluation, in using the K-BIT alone as the basis for decision making.

Intelligence is almost universally accepted as a construct involving a myriad number of contributing factors. Although the dichotomy of most intelligence tests used for screening purposes (including the K-BIT) provides an excellent summary of the components of IQ, they are essentially the most basic measures. Additional factors have their part to play in intelligence and intelligence testing, such as motivational, emotional and attitudinal variables. We must also consider the possibility that IQ as measured by intelligence testing is a cultural phenomenon, specific to our society. We must acknowledge that their are other facets to the construct of intelligence that we do not measure in basic screening tests such as the K-BIT. Consider creative productivity and artistic ability for example. Does the K-BIT account for the possibility of heritability and the decrement of intelligence with age, and to what degree? Not significantly, and that's why its recommended only as a screening device rather than a diagnostic tool.

A screening tests function is to identify individuals with scores above or below that expected for their age according to the standardized norm, and the K-BIT performs this task most effectively, but to stop there would be to dismiss qualitative investigation of the factors which may have lead to that result. The K-BIT therefore serves only to identify all those that require a follow-up investigation if scores differ significantly from the norm rather than provide an overview of that individuals intelligence in its entirety.

While similar to other screening tests in function and results, one of the most notable differences of the K-BIT in comparison with other tests of intelligence is demonstrated with the Weschler Adult Intelligence Scale - Revised version (WAIS-R). K-BIT scores are usually 5 points higher than WAIS-R scores, a difference which may be explain by the differences in administration between the two tests. Wechsler (1958, cited by Naugle, Chelune, & Tucker, 1993) theorized that the elements of intelligence were qualitatively different, leading to the formation of a verbal and visuo-spatial dichotomy which has carried into modern tests such as the K-BIT (evident in the Visual and Matrices subtests). What distinguishes the K-BIT from the WAIS-R is the fact that it relies less heavily on verbal output and is motor-free, which leads researchers to conclude that the 5 point difference may be due to differences in response efficiency.

Naugle, Chelune and Tucker (1993) have found that no significant evidence exists for the assertion that motor functioning or expressive language function account for the difference in scores. The only significant evidence for the difference in WAIS-R and K-BIT scores suggests that psychomotor speed may account for only 1.7% of the comparison, which is hardly remarkable. As we might expect from these results Naugle, Chelune and Tucker (1993) make a point of noting that, at present, there is no means by which we can predict which patients will demonstrate the discrepancies outlined.

The assertion that response demands may be responsible for the differences is valid if we take into account the time taken to administer each test. The longer WAIS-R is an unacceptable measure when time constraints, physical handicap or a patients stamina are considered. It is this factor which seems the more likely culprit for the difference in obtained scores, the short time needed to administer the K-BIT having a positive influence on scores in that we preclude confounding variables such as boredom or fatigue.

References

Gregory, R. J. (1996) Psychological testing: History, principles and applications (2nd ed.). Needham Height, MA: Allyn & Bacon.
Kaufman, A. S., & Kaufman, N. L. (1990) Kaufman Brief Intelligence Test (Manual), Circular Pines, MN: American Guidance Service, Inc.
Naugle, R. I., Chelune, G. J., & Tucker, G. D. (1993). Validity of the Kaufman Brief Intelligence Test. Psychological Assessment, 5, 182-186.