On Latent Change Model Choice in Longitudinal Studies