Shared-environmental contributions to high cognitive ability

Abstract

Using a combined sample of adolescent twins, biological siblings, and adoptive siblings, we estimated and compared the differential shared-environmentality for high cognitive ability and the shared-environmental variance for the full range of ability during adolescence. Estimates obtained via multiple methods were in the neighborhood of 0.20, and suggest a modest effect of the shared environment on both high and full-range ability. We then examined the association of ability with three measures of the family environment in a subsample of adoptive siblings: parental occupational status, parental education, and disruptive life events. Only parental education showed significant (albeit modest) association with ability in both the biological and adoptive samples. We discuss these results in terms of the need for cognitive-development research to combine genetically sensitive designs and modern statistical methods with broad, thorough environmental measurement.

Figure 1. IQ Distribution in the full sample in Study 1

Note: The dashed lines denote the relative position of the full sample mean (U), the proband mean (P), and the co-sib means for the three kinship groups in the high-ability subsample: adoptive siblings (A), DZ twins and biological siblings (D), and MZ twins (M).

Figure 2. Barplots depicting mean FSIQ for adoptive (lighter bars) and biological (darker bars) offspring, at each level of the three environmental variables: occupational status (2a), parental education (2b), life events (2c)

Note: For ease of presentation, family life-events scores were grouped into three-point intervals. There were life-events scores in the biological subsample that were greater than 11, the maximum in the adoptive subsample. Error bars represent ± 1 SE.

Figure 2. Barplots depicting mean FSIQ for adoptive (lighter bars) and biological (darker bars) offspring, at each level of the three environmental variables: occupational status (2a), parental education (2b), life events (2c)

Note: For ease of presentation, family life-events scores were grouped into three-point intervals. There were life-events scores in the biological subsample that were greater than 11, the maximum in the adoptive subsample. Error bars represent ± 1 SE.

Figure 2. Barplots depicting mean FSIQ for adoptive (lighter bars) and biological (darker bars) offspring, at each level of the three environmental variables: occupational status (2a), parental education (2b), life events (2c)

Note: For ease of presentation, family life-events scores were grouped into three-point intervals. There were life-events scores in the biological subsample that were greater than 11, the maximum in the adoptive subsample. Error bars represent ± 1 SE.

Figure 3

Diagram depicting mediation of parental education by latent shared environment. As shown in equations, approximately 4% of the shared-environmental variance is accounted for parental education.