Implement 'Park test'-like approach to choose optimal variance function for the costs part of the cost-effectiveness model (see Manning & Mullahy 2001 for discussion)
This requires estimating a GLM model to get predictions (& residuals) on the raw scale, then estimating ln(y_i - yhat_i)^2 = lambda_0 + lambda_1 * ln(yhat_i) +vi to get an estimate of lambda (i.e. lambda_1), which is the exponent in the variance function
Implement 'Park test'-like approach to choose optimal variance function for the costs part of the cost-effectiveness model (see Manning & Mullahy 2001 for discussion)
This requires estimating a GLM model to get predictions (& residuals) on the raw scale, then estimating ln(y_i - yhat_i)^2 = lambda_0 + lambda_1 * ln(yhat_i) +vi to get an estimate of lambda (i.e. lambda_1), which is the exponent in the variance function