Rindermann & Thompson (2013). Ability rise in NAEP and narrowing ethnic gaps?. Intelligence.
Quote: Third, achievements of African Americans (Blacks) and Latin American Americans (Hispanics) rose more than those of European Americans (Whites) (similarly Bishop, 1989): dec Black =3.04, dec Hispan = 2.27 and dec White = 1.29 IQ, leading to more narrow gaps (White – Hispanic-difference from diff = 11.59 to 8.46 IQ, White –Black-difference from diff= 16.33 to 9.94 IQ). Ability distributions are today more overlapping than in the past. This US development could be compared to the developments in Europe where ability gaps between natives and immigrants from Arabian-Muslim and African countries have narrowed, but haven’t disappeared (te Nijenhuis, de Jong, Evers, & van der Flier, 2004). But there are three disappointments: White–Black-gap-narrowing stopped in the late eighties, as also found by other measures (highest educational degrees, Graduate Record Exam, SAT, ACT; Barton & Coley, 2010; Farron, 2010, pp. 293-295; Neal, 2006; Rushton & Jensen, 2006 ); the competence level of high ability groups has hardly risen; and school leaver (17-year olds) have hardly achieved any gains.
The problem be …
(a) The gap was at its maximal in the 1970s — as such, comparisons between the present day gap and the 1970s gap don’t summarize well the historic trend. Refer to 100 years of Testing Negro Intelligence for a comprehensive review of studies.
(b) To make an achievement test, cognitive ability test comparison — NAEP-IQ, not NAEP-ACH — one has to correct for test correlations as discussed by Sackett and Ellingson (2000) and as done by Sackett and Shen (2008) and by Chuck (2013).
(c) The NAEP MAIN doesn’t support the narrative that “the gap” narrowed. Below are the NAEP age 17 LTT (Math and Reading) gaps from the 1970s to 2012 and the NAEP grade 12 MAIN (Civics, Economics, Geography, Math, Reading, Science, History, Vocabulary, and Writing) gaps from 2009 to 2012. The NAEP LTT (Math + Reading) gaps have indeed narrowed to d= 0.78, which is equivalent to a composite (or NAEP-IQ) d-value of 0.85 (assuming a subtest correlation of 0.7), yet the NAEP main (Civics to Writing) average d-value remains at 0.95, which is equivalent to a composite (or NAEP-IQ) d-value of 1.12.
There are two possibilities here: (1) NAEP average main gaps are “inflated” by the inclusion of exotic tests such as Civics (compare, though, with ASVAB tests such as Mechanics) or (2) both NAEP LTT and NAEP Reading tests are “deflated” by a program of teaching to the tests. It’s difficult to adjudicate between these alternatives. In support of hypothesis (2): the NAEP reading, SAT reading correlation (r=0.74) is significantly lower than the NAEP math, SAT math correlation (r=.91) — see Moran et al. — and the B/W narrowing has primarily occurred in NAEP reading, not NAEP math, etc; it is quite possible, then, that NAEP reading tests have lost some of their g-loading. In support of hypothesis (1), a similar pattern of differences (reading<math<science) has been found in the ECLS (1998-2007) and was found by Sackett and Shen (2008).
It’s not clear the me, though, that reading achievement differences better index the gap than the composite of all achievement difference taken together.
(d) Importantly, the gaps are still much larger than what would be predicted by the maximal reasonable genetic hypothesis d, genetic g =1. Assume heritability of g, age 17 = 0.7 & r (g x achievement) = 0.7. The achievement gap would be, on the account of genetic g alone: 0.84 x 0.7 = d = 0.59.
So, the question is: Why is the gap still so large? On the most general level, the answer is differential inheritance (environmental and genetic). Hence both the NAEP Main and LTT show that school identified mixed race individuals who self ID as Black and White score intermediate to school identified monoracial individuals — see, for example, It Could be Culture, part I/II (The NAEP Black-Mixed-White gap).
Moran et al. NAEP 12th Grade Preparedness Re Establishing a Statistical Relationship between NAEP and SAT
Sackett, Paul R., and Winny Shen. “Subgroup differences on cognitive tests in contexts other than personnel selection.” Implications of Organizational Staffing and High Stakes 31 (2009): 323.
Sackett, Paul R., and Jill E. Ellingson. “THE EFFECTS OF FORMING MULTI‐PREDICTOR COMPOSITES ON GROUP DIFFERENCES AND ADVERSE IMPACT.” Personnel Psychology 50.3 (2006): 707-721