Introduction

The purpose of this chapter is to assist the practitioner in the use and interpretation of multilevel models in the context of human growth research. The basic concepts of multilevel modelling are discussed and illustrated using a practical example. For detailed technical statistical discussions of multilevel modeling the reader is directed elsewhere (Goldstein, 1995; Kreft and de Leeuw, 1998; Snijders and Bosker 1999).

As we know human physical growth is a highly regulated process. From conception to full maturity the change in size and shape is a continuous process. Many attempts have been made to find mathematical curves that can fit, and thus summarize, the process of human growth. There is considerable literature on the analysis of longitudinal growth data both for linear (Vandenberg and Falkner, 1965; Berkey and Reed, 1987) and non-linear (Jenss and Bayley, 1937; Preece and Baines, 1978) parametric models. Adjusting a mathematical model to a set of growth data is called growth curve fitting or growth modelling. Such growth models have had variable success in describing the pattern of human growth depending on the type of growth variable used, the precision of the measurement, the frequency and age range of the observations and the ability of the model to describe the growth curve (Karlberg, 1998). At the individual level what is required is a curve with relatively few variables, each capable of being interpreted in a biological meaningful way (Tanner, 1989).