The predictive and diagnostic validity of the ITPA
The present study investigated the long-term predictive validity of the ITPA subtests with measures of achievement based on a routinely administered standardized test. One hundred and seventy-three children from 417 Ss who had taken the WISC, WRAT, and ITPA in the 1973-1974 school year were located in the Aldine Independent School District, Houston, Texas. The subjects were 173 children who had taken the Stanford Achievement Test (SAT) during the four year period following the administration of the ITPA. Group I consisted of four groups of children from 275 sets of scores who had taken the SAT one, two, three, and four years later; with 55, 80, 86, and 54 Ss in each group, respectively. Group II consisted of the test scores from 173 Ss when in the fourth and fifth grades, on the average of 2 1/2 years later. All Ss were initially identified as to sex, ethnicity, diagnostic category and SES level (based on Warner's Index). The ITPA was found to have good predictive validity over time, but predicted somewhat better for children who resembled the normative sample than for learning disabled children, males, and Mexican-American children. Grammatic Closure (GC) was the best overall predictor (p 4 .01), but its relative importance diminished over time. Auditory Sequential Memory (ASM) was the second best single predictor (p < .01), and its importance increased over time. Auditory Association (AA) and Visual Sequential Memory were found to increase in relative importance over time. GC, AA, and ASM were the best predictors for normal children. And, VSM, and to a lesser extent, AA, were the best predictors for learning disabled children, although these relationships were not consistent and were certainly not as conclusive as those for normal children. No ITPA subtest was found to be highly related to Arithmetic Computation. GC was found to significantly contribute to prediction equations, in conjunction with the other ITPA subtests, when IQ was controlled for Word Study and Language (p < .05). And, ASM significantly (p < .05) added to the prediction equations for all measures of reading and language when IQ was controlled. The results indicated that the ITPA retained its predictive validity over time, but that it predicts more accurately for Ss who resemble the normative sample. The ITPA should be used judiciously, and in conjunction with other instruments, especially for children who differ from the normative sample. Early language training programs, and programs designed to facilitate visual processing skills should be implemented in the classroom in order to adequately prepare all children for academic endeavors in terms of prerequisite skills related to those skills measured by the ITPA which were found to specifically relate to achievement.