The following description of the DAS subtests includes qualitative
characteristics for g loadings, reliability, and specificity. The following
definitions and criteria were used in assessing each subtest. Additional
information is available elsewhere in this volume and in four valuable sources:
Carroll (1993), Elliott (1997), Flanagan, Genshaft, & Harrison (1997), and
McGrew & Flanagan (1998, pp. 14-25, 64-68, 71, 63-91, 92-128).
When describing the norms for each subtest, the terms usual and extended
means that the subtest is appropriate for the full range of ability at that age.
The term Out of level denotes ages at which the subtest is appropriate
for most, but not all children.
The g Loading refers to the subtestís loading on the first unrotated
factor or component in principle factor analysis. A subtest with a general
factor loading of .70 or greater was considered Good; a loading of .51 to .69
Fair; and a loading of .50 or lower Poor. These are the same criteria used in The
intelligence test desk reference (ITDR): Gf-Gc cross-battery assessment
(Kaufman, 1979, pp. 109-110; McGrew & Flanagan, 1998, pp. 64, 72). Estimates
of g loadings were taken from Tables 9.4 (p. 202), 9.7 (p. 204), and 9.11 (p.
206) of the DAS Introductory and Technical Handbook.
Reliability refers to the degree to which a test score is free from errors of
measurement. A subtestís reliability was considered High if it was greater
than or equal to .90, Medium if it was greater than .79 but less than .90, and
Low if it was below .80. (McGrew & Flanagan, 1998, p. 64). Subtest
reliabilities were found in the DAS Introductory and Technical Handbook Tables
8.1 and 8.2 (pp. 178-179).
A subtest has three types of variance: common variance (that which is
shared with other subtests in the battery); specific variance (that
portion of the subtestís variance that is reliable and unique to that
subtest); and error variance (equal to 1 minus the reliability coefficient). We
cannot interpret an ability supposedly measured by an individual subtest, unless
that subtest contains a reasonable amount of reliable specific variance
(specificity) and this specificity exceeds the error variance. We computed the
specificity for each subtest at each age by the following procedure. The shared
or common variance was first estimated by the squared multiple correlation
between the specified subtest and all other subtests in the battery. Subtracting
the reliability coefficient from the common or shared variance provided the
estimate of specific variance for each subtest. Specificity was considered Ample
if the value was equal to or above 25% of the total test variance and it
exceeded the error variance, Adequate if it met only one of the two criteria
noted for Ample, and Inadequate if it did not meet either of the two criteria
noted for Ample. Again we followed the criteria listed in McGrew & Flanagan
(1998, pp. 64-66).
The McGrew, Flanagan, and Ortiz Integrated Carroll/Cattell-Horn Gf-Gc
Cross-Battery Approach Gf-Gc classifications are those proposed by McGrew &
Flanagan (1998). See also Carroll (1993); Flanagan, Genshaft, & Harrison
(1997); Flanagan, McGrew, & Ortiz (2000); McGrew (1997); and Woodcock