ABSTRACT
The Developmental Neuropsychological Assessment – II (NEPSY-II) is a widely used assessment battery in pediatric settings, but its internal structure has not been adequately examined. This study employed a rational, empirical approach to examine the construct validity of 23 NEPSY-II subtest scores from children ages 7–12 (M = 9.99, SD = 2.76) in the NEPSY-II norming sample (N = 600; 50% girls). Competing higher-order models based on prior research, hypothesized NEPSY-II domains, and conceptual subtest classifications were evaluated via confirmatory factor analysis and a sequential approach to model comparisons. The results supported the multidimensionality of NEPSY-II subtests and the organization of subtests by hypothesized neuropsychological domains. The best fitting model included a general factor and four first-order factors. Factor loadings from the general factor to first-order factors were very strong. However, general factor loadings for most subtests were less than .50 (range = .21–.69, M = .44), and domain-specific effects for all subtests, independent of the general factor, were even lower (range = .00–.45, M = .44). Interestingly, all subtests demonstrated strong subtest-specific effects, but it is not clear what construct(s) the subtest-specific effects represent. Findings support NEPSY-II authors’ emphasis on subtest-level interpretations rather than composite-level interpretations and highlight that NEPSY-II subtest scores should be interpreted carefully and with caution.
Acknowledgments
This manuscript is based on a thesis submitted by the first author under the supervision of the second author in partial fulfillment of the requirements for a doctoral degree in school psychology. We thank thesis committee member Dr. Kristoffer Berlin for his helpful comments.
Disclosure statement
We thank Pearson, Inc., for providing data from the NEPSY-II standardization sample. Standardization data from the NEPSY, Second Edition (NEPSY-II). Copyright © 2007 NCS Pearson, Inc. Used with written permission. All rights reserved.
Notes
1 Parameter values reported in figures may differ slightly from what is reported in text due to rounding error.
2 These values are the same as the total standardized indirect effects from the general factor to subtests (from Model 11C).
3 These values are the same as the difference between the squared multiple correlation for each subtest from Model 11C and the squared total standardized indirect effects from the general factor to subtests (a.k.a., the percentage of general factor variance for each subtest).
4 These values are the same as the difference between the internal consistency reliability estimate reported in and the squared multiple correlation for each subtest from Model 11C.
5 Error variance can be obtained by subtracting internal consistency reliability values in from 1.0 and converting them to a percentage.
6 The lower-order factor, Memory and Learning, could be viewed as measuring a general factor, and vice versa, the general factor could be interpreted as representing Memory and Learning.