Standards for the growth and development of children are generally presented in the form of an average growth curve together with some measure of variability, such as the standard deviation or percentile rankings, about the curve. Such standards may be constructed from either cross-sectional, longitudinal or mixed-longitudinal data. Although the mixed-longitudinal approach has definite theoretical advantages, several problems dealing with correcting the resulting standards for possible cohort and/or time-of-measurement effects arise in practice. In particular, the strategy of linking together longitudinal sequences from cohorts overlapping in age presents the problem of smoothing the growth curve and the corresponding percentile curves at the points of juncture of neighboring cohorts. This paper describes a method for generating smoothed standard deviation curves as a function of age, correcting for cohort differences. The procedure is applied to data collected as part of the Nymegen Growth Study. Assuming normality, these standard deviations are then used to obtain the corresponding percentile curves and an example of a smoothed growth standard is presented along with the 3rd, l()'/l, 25,h, 50
van't Hof, Martin A. and Kowalski, Charles J.
"Construction of Growth Standards from Mixed-Longitudinal Data,"
4, Article 7.
Available at: https://digitalcommons.wayne.edu/humbiol/vol49/iss4/7