Methods in Human Growth Research (review)

The cover of this useful and timely book sports the distance and velocity growth curves of the son of the Compte de Montbeillard, and these curves are superimposed on an image of George Louis LeClerc, the Compte de Buffon. Montbeillard measured his son twice a year from the boy's birth to his 18th birthday, and Buffon analyzed and published the data (1777), but he did not draw the curves (Richard Scammon did that in 1927). Montbeillard's data and Buffon's analysis form the first longitudinal study of human growth. Today, data collection and analysis are still two of the essentials of research (the third is publication). In the field of human growth and development, several books ably present the methods of data collection, including one by Noel Cameron. This new collection, edited by Hauspie, Cameron, and Molinari, is perhaps the first conveniently published resource for methods of both data collection and analysis.

The first section (six chapters) of Methods in Human Growth Research presents the rationale for conducting growth studies. The chapters are "Why Study Child Growth and Maturation" (John H. Himes), "The Human Growth Curve: Distance, Velocity, and Acceleration" (Luciano Molinari and Theo Gasser), "Sampling for Growth Studies and Using Growth Data to Assess, Monitor, and Survey Disease in Epidemiological Settings" (Edward A. Frongillo), "Measuring Growth" and "Measuring Maturity" (both by Noël Cameron), and "Measuring Body Composition" (Babette Zemel and Elizabeth Barden). All these chapters are quite informative, and the three chapters on measuring growth, maturity, and body composition are masterful reviews. The maturity chapter would benefit from the inclusion of a few hand-wrist radiographs showing the changes in maturity stages.

The next three chapters present nonparametric and parametric methods for analyzing individual growth. These chapters are "Kernel Estimation, Shape- Invariant Modeling, and Structural Analysis" (T. Gasser, Daniel Gervini, and L. Molinari), "Parametric Models for Postnatal Growth" (Roland C. Hauspie and L. Molinari), and "Parameter Estimation in the Context of Nonlinear Longitudinal Growth Models" (R. Darrell Bock and Stephen H. C. du Toit). Each chapter is written by the experts and provides valuable reviews of these quantitative approaches.

The third section of the book offers three chapters on the analysis of growth for populations. The first is "Univariate and Bivariate Growth References" (E. A. Frongillo), the second is "Latent Variables and Structural Equation Models" (Gino Verleye, Marie-José Ireton, J. Cesar Carrillo, and R. C. Hauspie), and the third is "Multilevel Modeling" (Adam Baxter-Jones and Robert Mirwald). Human biologists may be most interested in these population methods, and these three chapters serve as both an introduction and a practical guide to the use of these sophisticated techniques. The proper analysis of latent variables or concepts (e.g., socioeconomic status, dietary quality, perception of body image) is especially important to biocultural studies of human growth.

The final section of the book offers three chapters on special topics: "Methods for the Study of the Genetics of Growth and Development" (Stefan A. Czerwinski and Bradford Towne), "Prediction" (N. Cameron), and "Ordinal Longitudinal Data Analysis" (Jeroen K. Vermunt and Jacques A. Hagenaars).

Methods in Human Growth Research is aimed at junior and senior researchers and seems well suited for any serious growth researcher. A grounding in descriptive and inferential statistics is required of readers, but statistical expertise is not essential. Each chapter presents enough background and rationale to bring the reader "up to speed" on the topic. At the same time, this is not a cookbook of statistical recipes to be applied without understanding. Still, most chapters provide enough tangible examples of the application of the techniques that even a novice user of, say, a computer package for structural analysis will be able to see how to apply the methods.