Construct Validity of the Naglieri Nonverbal Ability Test for Black and White Second Grade Students

Abstract

The construct validity of the Naglieri Nonverbal Ability Test (NNAT) was examined with a sample of 367 white and 52 black second grade students from a rural Florida school district. Average item performance for both samples was similar, with average pass rates of 43% and 52% for the black and white samples, respectively. The rank order difficulty of the items was also similar, with Spearman’s rho (r) = .93 (p < .000001). Mean point-biserial correlations were .35 and .39 for the black and white samples, respectively, further supporting the construct validity of the test for both groups. Some differential item functioning was indicated, with nonsignificant point-biserial correlations for the black sample on 12 of 38 items. Point-biserial correlations were significantly different (p < .05) between samples for 2 items. Point-biserial correlations for each item were also correlated with the standardized difference in the proportions of blacks and whites passing each item, a measure of the degree to which the black-white difference was related to each item’s gloading. A Pearson (r) correlation of .28 (p < .05) was obtained. The convergent validity of the NNAT with the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) was also examined with a subsample of 22 students, yielding a correlation (r) of .74 (p < .0001). For the total sample, although mean IQ scores for blacks (86.7) and whites (96.4) were significantly different (p < .01), the differences cannot be attributed in any significant degree to test bias, with the test measuring the same construct for both groups. However, this test may not be a robust measure of the g-factor, at least for the age level of this sample.