The paper compared concurrent methods, two moment methods and two characteristic curve methods on the linear growth model with vertical scaling. The main findings were that, in general, the concurrent method performed better than others. Small sample size significantly affects the results of estimation. The growth intercept and growth rate were mainly affected by model, test length, and sample size. The variances of them were also considered here.
1. It implemented 100 replications in simulation and then we got a bias and RMSE. I’m curious how we can calculate the results of table 2 and the ‘mean’ bias, RMSE.
2. Before conducting simulations, what is our expectation of those methods’ effect on estimation? It looks the author just showed the results.