A Note on Two Models of Sibling Resemblance

This note compares two structural equation models of educational and occupational resemblance among sibling pairs. Each model decomposes the regression of occupational status on schooling into a between-family regression of common factors and a pair of within-family regressions. In model 1, the within-family regressions are writhin-family factors, and in model 2, the within-family regressions are written in the total educatinal and occupational variables. The two models are equivalent when within-family regressions are the same for each member of the sibling pair. Otherwise, they are not equivalent, and the second model has an undesirable logical implication. Under certain conditions, either model may exhibit symptoms of near-underidentification, or it may be underidentified; this problem may be more likely to occur in model 2 than in model 1.