Cantor R M
Acta Genet Med Gemellol (Roma). 1983;32(3-4):161-207. doi: 10.1017/s0001566000005055.
Two methods have been studied for extending the half-sib model, which was developed by Nance and Corey [34] for the genetic analysis of univariate traits, to include the analysis of multivariate traits. The methods are adaptations of the Bock and Vandenberg procedure [4] and the form of confirmatory maximum likelihood factor analysis which was developed by Jöreskog and Sörbom for the LISREL IV program [22]. These methods were applied to sex-adjusted individual finger ridge count data from the offspring of monozygotic twins. The Bock and Vandenberg procedure was applied to the eigenvalues and eigenvectors from a nested analysis of variance on 30 balanced male twin kinships. The result was a matrix of pure genetic effects which was at least positive semidefinite, and therefore appropriate for factor analytic procedures. Principal components analysis revealed two substantial genetic factors, one with a strong impact on the ridge counts of all ten fingers, with the largest loadings on the three central fingers of each hand, and the other influencing the thumbs and fifth fingers with opposite signs. In both cases, the factor loadings of homologous fingers were nearly equal. Employing the Bock and Vandenberg procedure to analyze multivariate data from MZ twin kinship has both positive and negative features. Its greatest strength is that it is easy to program with the Nested, Matrix, and Factor Procedures of the Statistical Analysis System package [see 44]. Multivariate half-sib data on any traits can be quickly explored for genotypic associations with the availability of this package or others like it. The exploratory findings from this analysis, the LISREL analysis, and the Bock and Vandenberg analysis on MZ and DZ twin pairs, as reported by Nance et al [36] are in agreement. This attests to the validity of the results from the two procedures. Such strong agreement may not be the case for other genetic structures, however, and analysis of other sets of traits or analyses on simulated data should clarify the cases under which the procedures produce concordant results. A negative feature of the Bock and Vandenberg procedure is that it wastes much of the data taken from individuals who attend the Twin Clinic. Both male and female kinships are routinely ascertained, with equal frequencies, although in this analysis we have examined the results from male kinships only. In addition, a substantial number of the male kinships either do not meet the criterion of at least two individuals in each sibship, or else have larger sibships containing individuals who must be excluded from the analysis.(ABSTRACT TRUNCATED AT 400 WORDS)
