The internal consistency reliabilities for the subtests are lower than those for the GCA and the three clusters, as would be expected. The average internal consistency reliabilities range from a low of .70 for Picture Recognition to a high of .91 for both Pattern Construction and Speed of Information Processing. At the Preschool level, across 9 age groups, and including subtests that at certain ages are considered "out of level," the median internal consistency reliability was .79. For the School-Age subtests, across the 13 age groups, the median internal consistency reliability was .83.

For the Preschool batteries, the core subtests had an average mean reliability of .82 compared to the diagnostic subtests' average mean reliability of .79. Pattern Construction (r = .88) is the most reliable Preschool subtest. For the School-Age battery, the core subtests had an average mean reliability of .84 compared to the diagnostic subtests' average mean reliability of .81. The core Pattern Construction (r. = .91) and diagnostic Speed of Information Processing (r. = .91) are the most reliable School-Aged subtests.

The average standard errors of measurement (SE_m) in standard score points are 3.80 for the GCA, 5.14 for the Verbal clusters, 4.85 for the Nonverbal clusters, and 4.19 for the Spatial cluster. Thus you can place more confidence in scores based on the GCA than in those based on either the Verbal, Nonverbal, or Spatial scores. In addition, you can place more confidence in scores obtained from the Spatial cluster than in those obtained from the Nonverbal Clusters and the Verbal Clusters.

In the standardization sample, the stability of the DAS was assessed by having 393 individuals from four age groups (3:6-4:5, 5:0-6:3, 5:9-6:11, and 12:0-13:11) retested after an interval ranging from 2 to 7 weeks (M = 30 days; DAS Introductory and Technical Handbook, p. 184). In the four age groups, the stability coefficients corrected for restriction of range were, respectively, .90, .94, .89, and .93 for the GCA; .84, .89, .87, and .89 for the Verbal Clusters; .79, .86, .80, and .83 for the Nonverbal and Nonverbal Reasoning Clusters; and .79, and .90 for the Spatial Clusters. Thus, the DAS provides highly stable GCA and Cluster scores.

Stability coefficients for the DAS subtests ranged from a low of .38 for Recall of Objects-Delayed at ages 3 years 6 months to 4years 5 months to a high of .90 for Pattern Construction at ages 12 years 0 month to 13 years 11months.

The mean test-retest scores and standard deviations for the Verbal, Nonverbal, Spatial, and GCA for the four age groups are presented in the DAS Introductory and Technical Handbook. On average, from the first to the second testing, the GCA increased by 3.0 to 7.8 points, the Verbal cluster increased by 1.2 to 5.1 points, the Nonverbal from 3.3 to 6.6, and the Spatial from 4.7 to 7.6 points. Measures of Verbal ability were somewhat more stable and showed smaller practice-effect gains than both the Nonverbal and Spatial abilities. At the composite levels, across the Preschool and School-Age batteries, the Verbal cluster increased about 2 points at the Preschool level and about 4 points at the School-Age level. The nonverbal clusters increased somewhat more (4 and 6 points respectively for the two levels), while the Spatial cluster averaged a 6.2-point increase.

Studies are needed to evaluate the stability of the DAS with other samples, including preschoolers and adolescents, and over longer periods of time. Such research would be helpful in learning about how cognitive abilities on the DAS change and in interpreting changes in scores when students are re-evaluated.

When the DAS is administered a second time, within 2 to 7 weeks, children are likely to have greater gains on the Nonverbal and Spatial subtests than on the Verbal subtests. Similarly to those changes noted on the Wechsler tests, this may happen because children may be able to recall the (a) types of items they were administered the first time and (b) strategies they used to solve the problems. During the first administration, children may perceive the Nonverbal and Spatial subtests as more novel than the Verbal subtests. On retest, these items may become less novel and perhaps more a test of long-term memory and ability to apply previous learning sets than a test of adaptability and flexibility.

Large retest gains on the Nonverbal and Spatial Composites raise concerns when interpreting the results when the DAS is readministered after a period of only 2 to 7 weeks. For periods longer than 7 weeks, gains on retest are likely to be lower because practice effects tend to diminish over time, but this assumption needs to be verified and quantified through research. A gain on the retest may have nothing to do with increased ability per se, and may simply reflect exposure to the test materials or practice effects.

Carefully consider whether you want to use the DAS for repeated evaluations, especially if you plan to use the results obtained on the retest for placement, eligibility, or diagnostic decisions. If the time between testing is relatively short, strongly consider using another individually administered, well-standardized test of cognitive ability for the reexamination.

On the preschool battery, the largest changes were for Recall of Objects-Immediate (increases of 2.6 to 5.5 T-score points), whereas the smallest changes were for Copying (decrease of 1 to an increase of .6). On a short-interval readministration, Recall of Objects-Immediate becomes in essence Re-Recall of Objects. Drawing copies of geometric designs, on the other hand, is a skill thoroughly practiced by many preschool children and therefore probably does not benefit much from a little additional practice.

On the School-Age battery, the largest changes were, again, for Recall of Objects-Immediate subtest (increases of 6.2 to 9.0 T score points), whereas the smallest changes were for the Recognition of Pictures and Recall of Digits subtests (averaging 1.3 and 1.7 points respectively). Recognition of Pictures is an out-of-level subtest from ages 8:00 through 17:11. For the six core subtests, the average gain ranged from 1.9 points for Word Definitions to 4.9 points for Pattern Construction.