Selection Bias in Estimation of the Genetic Correlation

Monte Carlo simulation of data was used to study empirically from parent and offspring covariances, the effect of truncation selection for one or both traits of the parent on the estimate of the genetic correlation between two traits. Ten replicates of samples of size 1000 were generated for each of 1225 combinations of parameters: the genetic correlation, ra ; the environmental correlation, re ; and the two heritabilities. Large biases in estimates of r, result when selection is intense. The levels of re and r, are important in determining the magnitude of the bias for both one and two trait selection. The levels of heritability are negligible for one trait selection and are only of slight importance for two trait selection. Regression equations are given for predicting bias caused by selection.