This paper introduced various types of models to describe the longitudinal data in developmental research, including the linear and non-linear models. Real data analysis supported the use of the more complex non-linear model. Given the data points are sufficiently large enough, complex model is advised to contain more effects in the model, like within/between-subject variation. The results are reasonable and not surprised, as complex models tend to capture more information from the data, especially when data size and computation are not problems. It is no wonder that the linear models are the weakest one to describe the real data. It is always said that “the best model is the data itself”. For researchers, to make a decision among the competed models is really a big problem. Is it really so important to make a choice purely based on the statistically criterion? I think for the practical use, whether the results are sensible to be interpreted, and whether there is strong theoretical evidence, are the most essential consideration.