부산대학교 도서관

While heritability of the Wechsler Intelligence Scales for Children (WISC) subscales has been estimated by several studies, none have investigated differential heritability or environmentality. Analysing a US sample of around 300 pairs of identical and non-identical 7 year old twins, we find significant differential heritability and/or environmentality across socioeconomic status (SES) for all subscales we examine except Digit Span and Coding. There has also been no previous investigation of how the genetic and environmental contributions to the relationship between intelligence and achievement change across SES. Using the same sample, we do not find any change in the components attributable to genes or to the environment of the covariance between intelligence (measured by WISC Full Scale IQ) and achievement (measured by the Wide Range Achievement Test (WRAT) sub tests Spelling, Reading and Arithmetic) . However, we find that the component of t he variance of WRAT Spelling attributable to genes increases with increasing parental education, as does the component of the variance of WRAT Reading attributable to environmental factors not experienced by both twins in a pair, leading to decreasing correlations with increasing parental education for these subtests. The findings have implications for the use of intelligence tests to select into educational opportunities. They suggest that WISC Full Scale IQ cannot determine which children of less educated parents would achieve highest were they to move to an environment similar to that provided by more educated parents. They also indicate that Digit Span and Coding are the fairest subscales of the WISC in that genetic and environmental factors are responsible for differences between individuals to the same degree across SES. We also focus on the methodology needed to carry out the analyses, evaluating existing techniques for estimating differential heritability and environmentality, stressing some important interpretational issues, and presenting new extensions to the basic multilevel genetic model that allow us to investigate the questions of interest.