Fig. 1
Sibling Correlations for Behavioral Traits. This figure displays sibling correlations for five traits measured in a large sample of Swedish brother pairs born 1951–1970. All outcomes except years of schooling are measured at conscription, around the age of 18. For details on sample construction and variable definitions, see chapter 3 in Cesarini.10 The sample sizes vary by outcome, but the minimum number of pairs per sibling type is MZT = 1,154; DZT = 1,601; FST = 151,789; FSA = 1,033; HST = 4,880; HSA = 11,566; ADO = 643. Because the sample sizes are large, all correlation coefficients are precisely estimated. The standard error of a correlation coefficient estimated with N pairs of siblings is approximately \((1-{\hat{\rho }}^{2})/\sqrt{N}\). For example, the approximate standard error of the 72% estimate reported in the full text (full siblings reared apart) is \(2\times (1-{\hat{\rho }}_{{\rm{FSA}}}^{2})/\sqrt{N}=2\times (1-{0.359}^{2})/\sqrt{1033}\approx 0.054\).
