Equation 4 suggests an estimating equation for any parameter of the quantitative model, given the association test between and . Write explicitly as a function of some parameter in Table 1, treating all other parameters as fixed and known. For example, might be the variance of marker effects in the training sample , from which is the explained genetic variance of the marker panel. Alternatively might be the covariance between marker effects in the two samples , assuming fixed values for the explained variances, and so on. Equation 4 is the squared coefficient of the linear regression of on , scaled by its sampling variance. The sampling distribution of that coefficient is normal, with mean the square root of equation 4. Therefore applying normal theory an estimator is the solution to the equation(6)where is the observed association statistic. An approximate 95% confidence interval for is given by where is the solution ofand is the solution of
