Segregation analysis of family data for 15 tests of cognitive ability

Segregation analysis of 15 tests of cognitive ability administered to 894 families of European ancestry and 366 families of Japanese ancestry gave evidence for a major gene contributing to performance on three tests of spatial ability in both ethnic groups. There was no evidence of major genes segregating in three other tests of spatial ability, or in tests measuring verbal ability, perceptual speed, or memory. The results supported the hypothesis that superior performance on ETS Hidden Patterns (HP) and Raven's Progressive Matrices (PM) is due in part to an autosomal dominant gene that accounts for 32% of the phenotypic variation in HP and 37% of the variation in PM. Evidence for a sex-limited autosomal dominant gene was found for a modification of the Shepard-Metzler Mental Rotations (MR) test, and 48% of the phenotypic variation in MR could be accounted for by this gene.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB 34720 and Grant HD 06669 from the National Institute of Child Health and Human Development.     

Familial resemblance for the Identical Blocks Test of spatial ability: No evidence for X linkage

The Identical Blocks Test of spatial ability was administered to subsamples of the two largest ethnic groups tested in the Hawaii Family Study of Cognition—Americans of European ancestry (171 families) and Americans of Japanese ancestry (98 families). Results of a hierarchical multiple regression analysis of family data and correlational analyses provided no evidence to support the hypothesis that spatial ability is influenced by a major, X-linked, recessive gene. Thus it appears that recent failures to replicate the sex-linkage pattern obtained by Stafford (1961) are not due to differences in the tests employed. We suggest that alternative explanations should be sought for the well-known sex difference in spatial ability.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development. The senior author also acknowledges support by NIMH Grant MH-11167.     

Parent-offspring resemblance for specific cognitive abilities in Korea

Regressions of offspring on midparent value for tests of specific cognitive abilities in Korea were considerably higher than those for Americans of Japanese ancestry or Americans of European ancestry tested in Hawaii. This greater parent-offspring resemblance in Korea may be due to the particular method of test administration or to an increased genetic variance resulting from assortative mating. The pattern of parent-child correlations for three relatively pure tests of spatial ability and for the spatial factor did not conform to that of a sex-linked recessive character.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Evidence for a major gene influencing performance on a vocabulary test

Scores on a vocabulary test given to members of 1818 nuclear families of different ethnic backgrounds were subjected to admixture analysis, segregation analysis utilizing both the mixed model and the transmission probability model and linkage analysis with 16 polymorphic markers. The data fitted a commingled distribution better than a unimodal distribution. Tests of the hypothesis that a major gene was segregating were not rejected by mixed model segregation analysis or by transmission probability analysis when provision was made for contribution from factors other than a major gene. There was no detectable heterogeneity by ethnic group nor by mating type. These analyses provide consistent support that there is a major gene for a component of verbal ability measured by a vocabulary test, and the segregation pattern is consistent with the expected Mendelian ratios. When vocabulary scores were adjusted for covariates including years of education, academic achievement, and reading habits, evidence for a major locus was lacking. The results of the linkage analysis were inconclusive. Current segregation models are affected by various factors leading to false inferences regarding monogenic mechanisms; however, in many respects, the mixed model and the transmission probability model are complementary with respect to power and robustness. The results of the analysis are discussed in this context.We thank Dr. J. M. Lalouel for access to his modified POINTER program, Professor N. E. Morton for access to NUCLEAR and POINTER, and Professor R. C. Elston for his support. The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado), supported by NSF Grant GB34720 and National Institute of Child Health and Human Development Grant HD06669.     

Blood polymorphisms and cognitive abilities

Data from 15 tests of cognitive ability obtained from the Hawaii Family Study of Cognition were examined for relationships with 18 blood polymorphisms. The number of significant associations did not exceed the number expected by chance alone. Significant regression of pooled verbal and spatial test scores on a constructed zygosity index demonstrated that increasing homozygosity was associated with improved scores. The effect was consistent in both parents and offspring, probably in both sexes, and for subjects of both Japanese and European ancestry. However, these blood polymorphisms contributed very little to total variation in test scores.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB 34720 and National Institute of Child Health and Human Development Grant HD 06669.     

The Hawaii family study of cognition: A reply

Differences in age and number of children tested per family between two ethnic groups (Americans of Japanese ancestry and Americans of European ancestry) have a trivial effect on differences in regressions of midchild on midparent for tests of specific cognitive abilities. Although the regression of offspring on midparent value is not a mathematical function of the spouse correlation, an empirical association between these two measures of familial resemblance is expected for characters with nonzero heritability. Such an association is reported for anthropometric characters. Since age effects contribute to the covariance of family members, age correction of family cognitive data is essential.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Familial resemblance for specific cognitive abilities

Measures of familial resemblance (spouse correlations, regressions of offspring on midparent, single-parent/single-child correlations, and sibling correlations) are presented for members of the two largest ethnic groups tested in the Hawaii Family Study of Cognition. Median spouse correlations (corrected for differences in test reliability) for 15 individual tests of specific cognitive abilities are 0.15 and 0.12 for Americans of European and of Japanese ancestry, respectively. With regard to the regressions of offspring on midparent value, corresponding median values are 0.50 and 0.35. Results of hierarchical multiple regression analyses, as well as the ordering of single-parent/single-child and sibling correlations, provide no evidence to support the hypothesis that spatial ability is influenced by a sex-linked, recessive gene.The results reported here are made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Familial resemblance for general cognitive ability in the Hawaii family study of cognition

Familial correlations for general cognitive ability wer estimated for three ethnic groups tested in the Hawaii Family Study of Cognition (Americans of European, Japanese, and Chinese ancestry). Assuming multivariate normality of family data, separate observed covariance matrices were computed for families of different size, ethnicity, and child's gender and then equated to their expectations using a log-likelihood ratio statistic. Resulting maximum-likelihood estimates of familial correlations are highly similar to previously reported product-moment and intraclass correlations; however, standard errors for the maximum-likelihood estimates are somewhat smaller. More importantly, the estimation procedure employed in the present study facilitates tests of alternative hypotheses. Using this approach, hypotheses of homogeneity of familial correlations among the three ethnic groups and of homogeneity among the various parent-offspring correlations (mother-son, mother-daughter, father-son, father-daughter) and sibling correlations (son-daughter, son-son, and daughter-daughter) were rejected.The results reported here were made possible by the collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and NICHD Grant HD-06669. This analysis was conducted while Beth Bennett was supported by NIMH Training Grant MH-16880.     

Sex-by-generation and ethnic group-by-generation interactions in the Hawaii Family Study of Cognition

Principal component scores for members of the three largest ethnic groups tested in the Hawaii Family Study of Cognition were subjected to a 3×2×2 (ethnic group x sex x generation) analysis of variance with unequal subclass numbers. Large and highly significant sex-by-generation and ethnic group-by-generation interactions were found for measures of general and specific cognitive abilities. These results suggest the influence of cultural factors on these group differences in Hawaii.The results reported here were made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by National Science Foundation Grant GB-34720 and National Institute of Child Health and Human Development Grant HD-06669.     

Correlations of measures of personality and of cognitive abilities within and across generations

Members of over 100 families took a battery of cognitive tests as well as various psychometric tests of personality. Analyses showed that individuals' cognitive and personality measures frequently were significantly correlated. Parent personality scale scores were significantly correlated with offspring cognition scores, and vice versa. Multiple regressions demonstrated that fathers' personality test scores substantially increased the predictability of sons' and, to a lesser extent, daughters' cognitive scores.The results reported here were made possible by the collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB 34720 and by National Institute of Child Health and Human Development Grant MH 06669.     

Handedness: An alternative hypothesis

Handedness data collected during the Hawaii Family Study of Cognition were congruent with similar data from major published studies. Mixed-model segregation analysis did not detect a significant major gene or polygenes contributing to handedness and partitioned the total phenotypic variation into 10–20% genetic and 80–90% environmental components. No relationship between handedness and cognitive ability was detected, but significant relationships between birth stress and offspring handedness were found. There was a significant decline in the percentage of left-handed individuals grouped by advancing age. A hypothesis of handedness is proposed in which left-handedness results from a combination of genotype, birth experience, and maternal example.The results reported here were made possible by collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by National Science Foundation Grant GB 34720 and National Institute of Child Health and Human Development Grant HD 06669.     

A reply to heathet al. on assortative mating for educational level

Heathet al. (Behav. Genet. 15:349–369, 1985), using data from a Norwegian twin registry and selected British and American samples, contend that assortative mating for educational level has not declined over the past 35 years. This is in contrast to the findings of Johnsonet al. Behav. Genet. 10:1–8, 1980), who reviewed the assortative mating literature, and Ahernet al. (Behav. Genet. 13:95–98, 1983), who analyzed data from the Hawaii Family Study of Cognition (HFSC). These authors found a decline in assortative mating for both intelligence and education. Heath and coworkers' criticism of the reliability of HFSC measures is rebutted, then new analyses are presented to show that the decline in assortative mating for both itelligence and education can be detected in self-report data from HFSC parents of Japanese ancestry, although not for parents of Caucasian ancestry. Data from a large-scale survey of alcohol use in Hawaii also indicate a decline in assortative mating for education. The differences in secular changes in assortative mating for educational level in Norway and in Hawaii appear to be real, not artifacts of measurement.The results reported here were made possible by the collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by National Science Foundation Grant GB-34720 and National Institute of Child Health and Human Development Grant HD-06669.     

Family resemblances in personality

Parent-offspring regressions, sibling correlations, and other measures of family resemblance in personality were computed for 54 personality traits, assessed by several psychometric personality tests as part of the Hawaii Family Study of Cognition. Considering all scales, the results suggest that a small familial/heritable component exists with regard to individual differences in personality, but did not demonstrate that some domains of personality are more familial or heritable than others.The results reported here were made possible by the collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB 34720, National Institute of Child Health and Human Development Grant MH 06669, National Institute on Alcoholism and Alcohol Addiction Grant AA 01762. The present report is based on a dissertation presented by the first author in partial fulfillment of the requirements for the Ph.D. degree.     

Family resemblances in serum uric acid level

There is substantial relation between individuals' serum uric acid level and their heights and weights. Even when the association of uric acid level with height/weight is partialed out, significant resemblances in uric acid level are found between biologically related persons (parents and offspring; siblings) but not between spouses. When the association of height/weight is partialed out, uric acid level does not have a significant association with measures of cognitive ability or with educational or occupational attainment.The results reported here are made possible by a collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Phenotypic assortative mating vs. social homogamy among Japanese and Chinese parents in the Hawaii Family Study of Cognition

Among the subjects of Japanese and Chinese ancestries in the parent generation in the Hawaii Family Study of Cognition were 47 pairs of siblings. Since data were available on the spouses of these siblings, this allowed for tests of whether spouse correlations of educational and occupational attainment and cognitive abilities were due to active phenotypic assortment and/or shared social background (social homogamy). Comparisons of sibling correlations, spouse correlations, and correlations between the spouse of one sibling and the spouse of the other sibling, as well as the results of model-fitting analyses, suggest that spouse correlations for education are determined by both phenotypic assortment and social homogamy, spouse correlations for occupational attainment by phenotypic assortment, and spouse correlations for verbal ability mostly by social homogamy.The results reported here were made possible by a collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii and J. C. DeFries, G.E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Personality attributes of males and females marrying within vs. Across racial/ethnic groups

As part of a larger study, Adjective Check List scores were obtained from 543 couples in Hawaii who had married within their own ethnic group and 83 who had not. Females who married across ethnic groups generally scored higher in one domain of personality (dominance) than did females who married within their own ethnic group. Males who married crossethnically also differed in dominance from those who did not, but in different directions in different ethnic groups. In the group that had the highest status (Caucasian) in the decades in which these subjects married, they were less dominant; in the other groups, to varying degrees, they were more dominant. In general, males and females in cross-ethnic matings were more similar in personality test scores than males and females who married within their own ethnic groups. Spouse correlations were generally positive but trivial in magnitude, with very little difference in personality between homogamous and heterogamous couples. The average difference in personality between those who married within and outside their own ethnic group was associated with the rarity of outmarriage for the sex and ethnic group concerned.The results reported here are made possible by the collaboration of a group of investigators (G. C. Ashton, R. C. Johnson, M. P. Mi, and M. N. Rashad at the University of Hawaii, and J. C. DeFries, G. E. McClearn, S. G. Vandenberg, and J. R. Wilson at the University of Colorado) supported by NSF Grant GB-34720 and Grant HD-06669 from the National Institute of Child Health and Human Development.     

Genetic models of reading disability

Test data collected on 133 reading-disabled (RD) children and their nuclear families who participated in the Colorado Family Reading Study were subjected to segregation analysis utilizing the technique of Elston and Yelverton (1975) for a continuous phenotypic measure. The possibility of genetic heterogeneity of RD was investigated by analyzing four subsets of data: all families, families with male probands, families with female probands, and families with severely affected probands. Furthermore, an analysis of the children's data was compared to that of all family members to investigate the possibility that the disorder may be manifested differently in adults. Results from the four subsets of data show that RD is etiologically heterogeneous. Compatibility with a major recessive gene for RD was demonstrated for families with female probands. Analyses of the children's data alone give results consistent with both environmental and genetic determination of RD.This word was supported in part by a grant from the Spencer Foundation, by NIMH Training Grant MH-11167, by NHLBI Training Grant HLO-599804, and by NIGMS Research Grant GM-16697.     

Commingling and segregation analysis of blood pressure in consanguineous and nonconsanguineous families from Andhra Pradesh,India

The hypothesis of a major gene effect for systolic blood pressure (SBP) and diastolic blood pressure (DBP) was examined in families from Andhra Pradesh, India, where consanguinity and marriage within caste are commonly practiced. The data were examined separately by noninbred versus complete (inbred + noninbred) families, both before and after correction for residual skewness. For SBP in the noninbred sample prior to skewness transformation, evidence consistent with a relatively common major recessive gene (q = 0.31) accounted for approximately 30% of the variance. In the combined sample, although Mendelian τs were rejected, the major gene estimates were similar to those in the noninbred data, and failure to account for inbreeding in the likelihood function may have influenced the results. After transformation of the data for residual skewness only a multifactorial component resulted, which accounted for about 80% (complete sample) and 100% (noninbred sample) of the offspring variance, and less than 10% of the parental variance. Even though the major gene effect disappeared after skewness transformation, the putative recessive major gene found for SBP prior to the transformation may be genuine since the tests on the transmission probabilities supported Mendelian transmission while the equal τs hypothesis was rejected. For DBP, both a major nontransmissible effect accounting for about 20% of the variance and a multifactorial component accounting for about 55% (offspring) and about 15% (parents) was found. For the noninbred DBP data, transformation for residual skewness removed the major non-Mendelian effect altogether. These data suggest inbreeding effects for SBP, with a pattern of smaller variances due to multifactorial sources (i.e., polygenic and/or cultural) in the sample which included consanguineous families. Adding inbred families reduced the evidence for Mendelian transmission of the major effect. © 1992 Wiley-Liss, Inc.     

Mixed-model segregation analysis of schizophrenia in the lindelius Swedish pedigrees

To test if familial transmission of schizophrenia is consistent with a model of monogenic inheritance with a multifactorial background, a mixed-model segregation analysis was applied to Swedish pedigrees consisting of 270 probands in 263 nuclear families. Results of the best-fitting mixed-model solutions are consistent with multifactorial transmission and no major gene. However, numerical instabilities prevented formal hypothesis testing, so an irrefutable genetic mechanism remains unidentified. Alternative research strategies that exploit recent advances in molecular genetics are discussed.Supported by grants from the John D. and Catherine T. MacArthur Foundation to Washington University, NIH Grant GM 28719, NIMH Grant MH 17104, and the National Alliance for Research on Schizophrenia and Depression.     

The electroencephalogram in the prediction of human reaction time during growth and development.

Characteristics of the distributions of electroencephalographic (EEG) half waves recorded in children during performance of a simple auditory reaction time (RT) task were investigated. The purpose was to determine the extent to which difference in these distributions could account for children's slow RT. The durations of a sample of 760 EEG half waves in each of a group of 41 healthy children aged 5-17 yr were measured and distributed into an interval histogram, and the first four central moments of the 41 distributions were computed. All four of the moments-which measure the central tendency, dispersion, skewness and kurtosis of the distributions-proved to be significantly correlated (p less than 0.01) with RT. The multiple correlation (R) between RT and the four moments was equal to 0.68 and was statistically significant. R was unchanged when only the measures of dispersion, skewness and kurtosis were used as predictor variables, suggesting that the predictive value of the central tendency is low. The predictive capability of a multiple regression equation based on these three predictor variables was tested in an independent group of 42 children also aged 5-17 yr. EEGs and RTs were recorded and this group was treated in exactly the same way as the other group. The correlation between these children's average RTs and their RTs as estimated from the other group's regression equation was 0.53. These findings suggested that knowing the degree of dispersion, skewness and kurtosis of the children's EEG distributions reduces from chance by about 28% the error of predicting their RTs.