Thursday, September 24, 2009

RIAS concurrent validity study with WJ III in university population

Krach, S.K., Loe, S. A., Jones, W. P. & Farralay, A. (2009).  Convergent Validity of the Reynolds Intellectual Assessment Scales (RIAS) Using the Woodcock–Johnson Tests of Cognitive Ability,Third Edition (WJ-III) With University Students.  Journal of Psychoeducational Assessment, 27 (5), 355-265.

Validity studies with the Reynolds Intellectual Ability scales (RIAS) indicated that RIAS composite intelligence index (CIX) and verbal intelligence index (VIX) scores have moderate to- high correlation with comparable scores on other instruments. The authors of the RIAS described the VIX scale as a measure of crystallized ability and the nonverbal index (NIX) as a measure of fluid ability, but no studies were available comparing the VIX and NIX scores with an established measure of these abilities, nor studies specifically addressing the utility of the RIAS with a university population. The study examined RIAS scores of university students in comparison to scores on the Woodcock–Johnson Tests of Cognitive Ability (third edition; WJ-III). Consistent with previous studies, there were moderate-to-high correlations between the CIX and VIX scores and corresponding scores on the WJ-III. A substantially lower correlation was evident between the NIX and WJ-III fluid ability scores.
Article Highlights
  • The sample consisted of 107 undergraduate university volunteer subjects.  The authors point out that this is the first RIAS concurrent validity study in a univeristy population and also the first study to use an established validated composite index of fluid intelligence (Gf; WJ III Fluid Reasoning cluster) from an independent intelligence battery (WJ III).
  • As would be expected given the highly selective sample, the range of talent in the study was restricted (SD's ranged from approx 9 to 10-11 standard score points).  As a result, the author reported both the obtained and range-restricted corrected correlations between the three RIAS composite scores and the WJ III Tests of Cognitive Abilities GIA and Gc and Gf cluster scores [conflict of interest note - I'm a coauthor of the WJ III].
  • The primary correlations of interest are those for the observed/corrected overall composite IQ scores (.57/.75), the RIAS Nonverbal (NIX) and WJ III Fluid Reasoning (Gf) scores (.39/.54), and the RIAS Verbal (VIX) and WJ III Comprehension-Knowledge (Gc) scores (.72/.88).
  • Based on the above corrected RIAS NIX/WJ III Gf correlation (.54), the authors concluded that "the data in this study do not appear to support the use of the NIX scores as interpretable under the Gf–Gc or CHC frameworks. Further study is needed to determine whether this outcome is a result of the design of the NIX scale, a limitation of the WJ-III fluid ability measure, or an artifact of this sample."
  • The authors also reported that "the composite scores from the RIAS were significantly higher than those from the WJ-III."  The authors reported that the RIAS composite scores were approximately 3-4 points higher than the WJ III GIA score.  In important qualification (by the blogmaster) is that the authors used the original WJ III 2001 norms and not the later WJ III Normative Update norms, which in general, provide slightly lower scores for the WJ III...which would most likely make the RIAS/WJ III global composite score differences slightly larger.
  • It would have been nice if the study would have reported the intercorrelations among all tests so independent researchers could examine the pattern of correlations and possibly import them into SEM programs and run a quick CHC-driven CFA study.  More joint or cross-battery intelligence studies, like this one, are sorely needed.  There was a wave of them a decade or more ago...but they now are few and far between.
  • Although the corrected global composite IQ correlation is statistically significant and "high" (.75), correlations of this magnitude should not be interpreted as indicating that each test will provide nearly identifical general intelligence scores.  This issue is relevant to all IQ/IQ test comparison studies and is not specific to this study.  This was recently discussed and illustrated in an IAP Applied Psychometrics 101 research report.
  • As the authors acknowledge, caution is in order in making grand generalizations from this study given the highly selective nature and characteristics (university undergraduate volunteers) of the sample.
Technorati Tags: , , , , , , , , , , , , ,

No comments: