Open MorphoFun opened 8 years ago
estimates for the nonlinear terms improve a little bit if I change the correction to mean(phat(1-phat)^2); need to confirm whether this is a good modification
The methodology for the Janzen and Stern (1998) approach was originally intended for only estimating linear selection gradients. It does not seem to apply to nonlinear selection gradients, so it might be best to create a separate function to conduct the Janzen and Stern (1998) approach for linear selection gradients only.
estimates from nonlinear terms for JS do not match OLS