Classification Using Growth Curves

Abstract

Classification using the growth curve model is analyzed according to whether the covariance matrix structure is known to be compound symmetric. A new classification procedure appropriate for classification in the case of models with means that follow a Potthoff and Roy [Potthoff, R. F., Roy, S. N. (1964). A generalized multivariate analysis of variance model useful especially for growth curve problems. Biometrika 51: 313–326] structure is suggested. Simulations are carried out to illustrate this new procedure and to estimate the chance of misclassification.

Key Words:

• Growth curve
• Classification scores
• Model-based covariates
• Simulation

Statistics Subject Classification:

• Multivariate Analysis—Inference

Acknowledgments

The authors want to thank the referee and the associate editor for their helpful comments, which improved the original version of this article.