Introduction

Prediction within the context of research in human growth and development may be defined as the estimation of the magnitude of some unmeasured dimension of interest. Prediction may take place in both cross-sectional and longitudinal scenarios. In cross-sectional scenarios one may wish to predict a dimension that cannot, for whatever reason, be measured at the time of assessment of other dimensions. For instance, one may wish to estimate total body fat when it is not possible to take a direct measure but when anthropometric estimations of subcutaneous fat (skinfolds) are available. In longitudinal situations one is interested in predicting the magnitude and/or timing of some future event, such as birthweight, or the timing of menarche, or adult stature.

The process of human growth and development lends itself particularly well to prediction methods for two reasons; first, almost all physical dimensions are associated with each other and do not vary independently, and second, human somatic growth is a process of change over time and has, for the most part, distinct and measurable end points. The end points also do not usually vary independently of earlier size and shape. However, the magnitude and timing of human growth and development are also variable and thus whilst it may be known with certainty that a particular event will occur, e.g. the achievement of adult stature, it is not known precisely at what time the event will occur or what magnitude it might be.