Normed for ages 5:0 to 17:11 (Usual age range is 6:0 to 17:11; Out of level age range is 5:0 to 5:11).

This subtest assesses acquired verbal knowledge and language comprehension and fluency as well as verbal inductive reasoning; vocabulary and verbal development; logical and abstract thinking; and ability to distinguish between essential and superficial features.

The Similarities subtest contains a total of 34 three-word items. Children are read three words and asked to tell how they go together, what they all are, or how they are similar. Responses are typically scored 1, or 0, although 6 items do provide the option for a 2-point response as well. Three different starting points are available (age 5:0 to 6:11 start at item 1; 7:0 to 8:11 start at item 8; 9:0 to 17:11 start at item 13).

The Similarities subtest is considered a fair measure of g across all ages (overall r = .69). For ages 6:0 to 12:11, the g loading is considered fair, while at the remaining upper ages (13:0 to 17:11) the g loading is considered good. This subtest contributes substantially to the Verbal factor (loading = .81). Specificity is ample for all age groups 6:0 to 15:11 and adequate for 16:0 to 17:11.

Similarities is considered to possess low overall reliability (r = .79), with reliability coefficients ranging from .73 to .84 across all of the 13 whole-age groups. There may be a certain amount of chance involved in trying to think of a correct link among the words instead of a no-credit, trivial, but also true connection. Similarities correlates best with Word Definitions (r =.64) and least with Recall of Objects - Delayed (r = .15). It has a moderate correlation with the GCA (r = .75).

Gf-Gc classification

In the Broad stratum definition of abilities, Similarities is considered to be a strong measure of Crystallized Intelligence (Gc). In the Narrow stratum of abilities, it is considered to be a probable measure of Language Development (LD) and a possible measure of Lexical Knowledge (VL) (McGrew & Flanagan, 1998, p. 124).

Similarities differs dramatically in a number of ways from the subtest with the same name on the Wechsler scales. The DAS provides three target words. This allows a child who may not know the meaning of one of the three to use the other two to develop a response. (How often has a child who did not understand the meaning for the word TRIBE received a 0 score on that WISC-III Similarities item?) The SB: FE Verbal Relations subtest asks the student to explain how three words are alike and different from a fourth. The DAS also differs from the Wechsler scales in that a child must respond with a superordinate class for the stimulus words in order to earn credit. On the Wechsler Scales, a subordinate response is given 1 point while the superordinate response is given 2 points, after the first five items. Hypothetically two children can obtain the exact raw score of 13 (and thus the same scaled score), one giving 13 one-point subordinate responses while the other simply gives 4 two-point superordinate responses after the first five items.

Again, the DAS calls for liberal questioning of responses. "Question a response that is incorrect but that indicates some understanding. . . . On 2-point items, also question all 1-point responses" (Elliott, 1990a, p. 229).

The Sample and the first two actual items administered are teaching items if the child gives an incorrect response. The examiner acknowledges correct responses to those items. The examiner may use any of four means of presenting items, including simply saying the three stimulus words once the child understands the task.