Because of the many issues associated with identification of gifted and talented children, especially with underrepresented populations (Passow & Frasier, 1996), Lohman (2009) argued for a different approach to identification, one based on academic aptitude. One of the major issues when attempting to identify talent in underrepresented populations is the aspect of norm referencing: "Those who do not understand the relativity of norms – especially on ability tests – miss the easiest and most effective way to identify minority students who are most likely to develop academic excellence" (Lohman, p. 976). Too often school districts and others rely on nationally normed referenced tests when, in actuality, local norms are much more appropriate (p. 975). Lohman suggested an identification method that used ability test scores with multiple norms, including local and even subgroups within the local populations (e.g., English Language Learners). Those scores would then be combined for one verbal/reasoning score and one mathematical/quantitative score. In addition to those scores, teacher ratings would be incorporated and compared to the scores. Aptitude would then be based on the group itself, and greater identification of underrepresented populations would occur.
In determining an ideal protocol for identification, Project GEMS followed Lohman's suggestions. Local norms were used on the math and science subscores of the Iowa Test of Basic Skills (ITBS), Form C (Hoover, Dunbar, & Frisbie, 2007). Students were selected based on their relative scores to others in their grade and school. In addition to the ITBS, the nonverbal subscore from the Cognitive Abilities Test (CogAT; Lohman & Hagen, 2001a) was used. As discussed by Lohman (2008), teacher input was also a factor. The Project GEMS Evaluator created Teacher Identification Form: Science and Teacher Identification Form: Math to be part of the identification protocol. All standard scores were transformed into z-scores based on means and standard deviations across the six schools for each grade. Z-scores were calculated for the five measures individually. Composite z-scores based on the combined five measures were used to select students. The composite scores for overall identification (i.e., ITBS math, ITBS science, CogAT, Teacher Identification Form: Science, and Teacher Identification Form: Math) were used to identify the top 25 students in each grade (i.e., second through fifth) at each of the six elementary schools in Spring 2009 so that students could begin treatment in Fall 2009 as third through sixth graders.
The ITBS has a long-standing reputation of being the "most venerable standardized, norm-referenced achievement test batteries in existence today" (Engelhard, 2007, para 18). The total composite measure has high internal consistency coefficients (r = middle .80s to low .90s) based on the Kuder-Richardson Formula 20 (Engelhard, 2007; Lane, 2007). Equivalent-forms reliability between Forms A and B are high with correlations in the high .70s and .80s (Lane, 2007). Based on the information from the 2000 national standardization, the stability coefficients are equally high (Lane, 2007). As to validity, the authors of the test emphasize the importance of item-by-item examination by schools, districts, and states to establish curricular relevance (Engelhard, 2007). The process used to design the test followed the national standards for test design (Lane, 2007). Correlations among subtests and composite scores are moderate to high in regard to internal validity, and standard errors of measure are similar in regard to gender and ethnicity (Lane, 2007).
Likewise, the psychometrics for the CogAT are strong. It was standardized in 2000 in conjunction with the ITBS (Lohman & Hagan, 2001b). Lohman (2008) found the CogAT to be more reliable than several other measures assessing aptitude, including the Wechsler Intelligence Score for Children (WISC-IV), Otis-Lennon School Ability Test, and the Naglieri Nonverbal Ability Test. As to construct validity, there is a strong correlation between the CogAT composite (including the nonverbal section) and WISC-IV full-scale score which includes Perceptual Reasoning (r =.79) (Lohman, 2008). Lohman argued that the measure is "excellent for predicting current and future academic achievement" (slide 54) because of the strong within-battery predictions (i.e., nonverbal with math r = .4 to .7); moreover, he found the predictions to be the same for all ethnicities. Teacher Measures Psychometrics for the Project GEMS-generated Teacher Identification Form: Math and Teacher Identification Form: Science are also strong (Roberts, Inman, Wininger, & Tassell, 2010). For the science measure, the overall reliability across the grades was .87 which indicates confidence in the internal consistency. Moreover, validity was established through moderate correlations between student scores on the ITBS science section and their ratings on the identification form; correlation coefficients were significant at the .01 level (third grade, r = .34; fourth grade, r = .41; and fifth grade, r = .38). Similar results were found on the math identification form. It had high internal consistency via coefficient alpha across the grades (α = .93). Correlation coefficients between the form and the math subsections of the ITBS were found to be significant at the .01 level (third grade, r = .38; fourth grade, r = .42; and fifth grade, r = .46). Both measures, then, demonstrate adequate reliability and validity.
Some of the links on this page may require additional software to view.
