Abstract
A growth function is a mathematical functional form that models longitudinal growth over time. Longitudinal growth is often difficult to model because of variation in the observed data, which makes it difficult to fit a growth curve. Thus, several different functional forms have been proposed as growth functions. Growth functions are typically selected based on how well they fit the data - final analysis is fully based on the selected function. If the wrong growth function is selected, results may be skewed or invalid, underscoring the importance of proper model selection. There are no definitive guidelines for selecting the most appropriate growth function from the set of available candidates. We propose a statistical approach for growth form selection that uses Mallows' Cp criterion.
